AI Applications

Finance has undergone a significant transformation in recent years, shifting from a static function to a real-time one. This change is not just about closing the books faster or refreshing forecasts more often; it’s about seeing the business as it changes and helping leaders act sooner. A finance team that can provide this level of insight can give the company more time to shape what happens next.

When I joined OpenAI two years ago, we had a small finance team supporting a rapidly growing company. We needed to build our function from scratch and make AI fundamental to how we work, make decisions, and support the business. This meant redesigning our processes around artificial intelligence tools that could help us analyze data more efficiently.

One of the key challenges we faced was manual, recurring work: finding information, explaining what changed, and assembling inputs for a decision. We had access to advanced AI tools but were still learning how to use them effectively. So, we set two ambitious goals: achieving a zero-day close and automated, continuously updated forecasting.

A zero-day close means giving leaders a real-time view of the company’s financial position, reconciled and traceable. Continuous forecasting builds on this foundation by showing how the business is changing, what could happen next, and which decisions can alter the outcome. We’re still working toward these goals, but our efforts have already changed how we operate.

Our team has moved beyond static spreadsheets, manual searches for supporting records, and presentations to live tools built on the full context and data of the business. Finance professionals now have the ability to build their own decision-making infrastructure using AI tools like ChatGPT Work and Codex.

The real promise of an AI-native finance function is a team that understands what’s happening as it happens, helps leaders see choices ahead, and gives the company more time to act while outcomes can still change. To get there, we need to redesign work around decisions that matter, give people room to experiment, build clear accountability into every workflow, and measure AI’s dependable work.

One of our key takeaways is the importance of broad access to AI tools. People need freedom to explore these capabilities in their own work context. We paired this with structured experimentation around real problems, which led to some innovative solutions like IR-GPT, a custom GPT for investor relations.

Our hackathon turned AI from an abstract capability into a working tool. In one day, people could identify recurring tasks, build solutions, test them with colleagues, and improve them. The use cases came from those closest to the work while technical experts helped move faster.

For CFOs, our lesson is simple: you need bottom-up experimentation combined with top-down strategy. Put secure AI in people’s hands and let those close to the work identify better ways of getting things done. At the same time, focus leadership attention on changes that will matter most to the business.

SFA athletics has made a significant move in the world of sports storytelling by partnering with FanWord, an AI-powered platform used by over 215 schools and conferences across the United States and Canada. The partnership aims to ‘humanize’ box scores and stats by focusing on the people behind them rather than just numbers themselves.

The multi-year deal gives SFA’s athletic communications team access to FanWord Assist, a suite of tools designed to turn box scores, stats, and postgame interviews into ready-to-edit game recaps, player bios, and award nominations. This AI tool is not only used by Stephen F. Austin but also by other prominent institutions such as Florida State, UConn, and SMU.

According to Jacob Pumphrey, SFA’s assistant athletic director for communications, the goal of this partnership is simple: tell better stories faster while still maintaining a human touch in sports reporting. ‘The human element is not going to be lost,’ he emphasized. ‘We still have to be good crafters of the written word. We still have to know how to storytell.’

Pumphrey explained that FanWord Assist will help his three-person communications office work more efficiently, freeing up time to pursue deeper and more meaningful athlete stories. He noted that this partnership is not a replacement for human journalists but rather an aid in streamlining the process of creating game recaps.

The program costs between $250 and $350 per sport, which Pumphrey believes is roughly an eighth of what a full-time staff position would cost. This makes FanWord Assist an attractive option for athletic departments looking to cut costs without sacrificing quality content.

What sets FanWord apart from other AI tools is its focus on the needs of people working in athletic communications. ‘The one thing that kind of set FanWord apart from just your normal AI is the fact that it’s a company that employs people who have actually worked in these positions,’ Pumphrey said. This understanding of the industry allows FanWord to craft tools tailored specifically to the challenges faced by those in this field.

One feature that Pumphrey is particularly excited about is the platform’s first-person storytelling tool, designed to help tell athlete stories even when schedules make sit-down interviews nearly impossible. ‘Student athletes’ schedules are very busy,’ he noted. ‘It’s tough to maybe find a 30-minute window sometimes to sit down and talk with them.’ This tool allows for text answers, audio answers, or video answers from student-athletes at their convenience.

Not everyone is convinced that this partnership is the right move, however. Kylie Chandler, a junior at SFA and writer for The Pine Log student newspaper, expressed disappointment over the decision to use AI in game recaps. ‘I feel like it’d be so much more meaningful if we used this opportunity to showcase our students’ abilities,’ she said.

Chandler also raised concerns about mixed messaging on campus regarding AI usage. She noted that while SFA is moving forward with AI-powered tools, students are discouraged from using AI in their coursework. Pumphrey pushed back against the idea that the partnership signals a lack of confidence in student journalists.

‘This isn’t a move because we don’t want to rely on our journalism department or we don’t feel that students can get the job done,’ he said. ‘This is just more so to help the current full-time employees in the office do their job a little bit more efficiently.’

The door remains open for student workers, with Pumphrey encouraging any interested students to come forward and discuss potential opportunities. The athletics department says it is actively looking for student workers and interns heading into the fall season.

Ultimately, this partnership marks a significant step towards ‘humanizing AI’ in sports storytelling, but its impact will depend on how well SFA’s athletic communications team can balance efficiency with the need to tell meaningful stories about athletes.

A standards-setting organization is set to address the role of autonomy in aviation at its upcoming symposium. ASTM International’s Autonomy in Aviation Symposium, scheduled for early October in Jacksonville, Florida, will focus on how advanced systems can reduce reliance on exceptional pilot skill or extensive training.

The 7th Annual ASTM Autonomy in Aviation Symposium aims to pave the way for increasingly capable automation. According to the event webpage, hosted by ASTM’s Autonomy Design and Operations in Aviation Administrative Committee, the goal is to safely move the industry towards an automation-centric design paradigm.

Wes Ryan, committee chair and Northrop Grumman fellow for airworthiness of autonomy and AI, emphasized that the focus is on allowing automation to handle ‘low-level’ events, such as basic system malfunctions. This would enable pilots to manage safe outcomes at a higher level, focusing on mission outcome rather than intervening in routine tasks.

ASTM creates standards for materials and methods that represent voluntary consensus in an industry. These standards are frequently adopted by regulators, including the FAA. For instance, last year’s MOSAIC rule updated light-sport aircraft and sport pilot privileges, citing four ASTM standards.

The autonomy committee is not yet tackling specific standards but intends to publish a paper on automation in aviation within a year after the symposium. Ryan noted that this will provide crucial guidance for industry stakeholders.

Since 2019, ASTM has published increasingly specific guidance on this topic. For example, its 2020 technical report aimed to promote the sound application of technical best practices and increase aviation autonomy. The group also highlighted unresolved questions around autonomous flight in a 2023 white paper.

The symposium is meant to foster conversation around these and other questions. Ryan emphasized that there is a need for increased industry consensus and regulatory guidance, given the start of FAA’s eIPP initiative to accelerate the rollout of electric aircraft services and the U.S. military’s growing adoption of autonomous aircraft and related technologies.

In particular, the October event will focus on best practices for aircraft design that ensure safety while allowing the role of the pilot or operator to adapt depending on the level of automation. Ryan views this as a shift away from requiring pilots to intervene in routine tasks and toward allowing automation to handle such events.

This approach would enable pilots to ‘manage safe outcomes at a high level, a mission outcome level,’ so they become more systems managers rather than being involved in low-level tasks. Ryan described this concept as ‘carefree handling.’

Speakers confirmed for the event include David Mindell, an aerospace professor at MIT and author of the book ‘Our Robots, Ourselves,’ along with representatives from FAA and NASA. Boeing-owned Wisk’s regulatory affairs lead Tom Gunnarson will also be in attendance.

Gunnarson noted that ASTM represents a broad forum to steer oversight of new technology. He emphasized that regulators are focused on safety, so if they indicate developing performance-based rules, standards bodies can develop the means of compliance.

The aviation and aerospace industries will closely watch the conference and general discussion on autonomy. Ella Atkins, a professor of aerospace who directs Virginia Tech’s Autonomous Aerospace Systems Lab, highlighted the importance of ‘trust calibration’ in autonomous systems.

Atkins explained that humans need to be able to trust autonomy while also having override capabilities if the machine starts to malfunction. Unfortunately, flight crews may not always have the best intentions, so machines must ensure their trust in humans is not absolute.

Tuskegee University has been awarded a significant research grant from the U.S. National Science Foundation (NSF) to advance trustworthy artificial intelligence (AI) in healthcare settings. The three-year project aims to improve the safety, transparency, and reliability of AI-assisted medical technologies by developing more robust systems that can generate responses grounded in reliable evidence.

The initiative is housed within the Department of Computer Science in the Andrew F. Brimmer College of Business and Information Science at Tuskegee University. Led by Assistant Professor Dr. Baiyun Chen as principal investigator, the research team will combine evidence-based reasoning with compliance-aware AI methodologies to support clinical guidelines and regulatory requirements.

One key aspect of this project is its focus on incorporating meaningful human oversight into high-risk healthcare decision-making processes. By doing so, the researchers hope to increase confidence in AI technologies used across various healthcare settings. This approach also acknowledges that AI assistants can only be as effective as their ability to work alongside humans.

The research team will draw expertise from multiple disciplines, including artificial intelligence, cybersecurity, and healthcare informatics. A notable collaboration is with the University of North Texas, which brings additional knowledge in these areas. The project’s scope encompasses not just technical advancements but also workforce development and faculty mentorship opportunities for students.

According to Dr. Thierno Thiam, Provost and Senior Vice President for Academic Affairs at Tuskegee University, this grant represents a significant milestone in the institution’s progress as leaders in shaping the future of AI in higher education. He notes that beyond immediate impacts on trustworthy AI development, the grant strengthens research capacity and reinforces the university’s commitment to its mission.

The project also highlights the importance of interdisciplinary collaboration in addressing real-world challenges. Dr. Sean Edmund Rogers, Dean of the Andrew F. Brimmer College of Business and Information Science, emphasizes Tuskegee University’s unique position in bringing together computer science and business scholars to explore discoveries that will shape future work practices.

Dr. Fan Wu, Department Head of Computer Science at Tuskegee University, sees this project as a meaningful opportunity for faculty-led research while providing students with hands-on experience tackling pressing AI challenges in healthcare. Through this initiative, the university aims to build the next generation of researchers and innovators equipped to develop responsible AI solutions.

The NSF award not only advances research but also expands opportunities for undergraduate and graduate students to engage in interdisciplinary projects at the intersection of artificial intelligence, cybersecurity, healthcare, and data science. This grant further reinforces Tuskegee University’s growing leadership in these areas while supporting its mission of advancing knowledge through education, service, and research.

The project is titled ‘Excellence in Research: Trustworthy AI for Healthcare: Evidence-Grounded, Compliant, Human-in-the-Loop,’ reflecting the comprehensive approach it takes to address healthcare challenges with trustworthy AI. This initiative underscores Tuskegee University’s commitment to innovation and its role as a leader in shaping the future of AI applications.

Electric vehicle giant Tesla has finished dismantling its original assembly line for the flagship Model S sedan and Model X SUV at its Fremont factory. The decommissioning process wrapped up in just 46 days, which is a remarkably short time frame considering the complexity of the operation.

The Model S, introduced in 2012, and the Model X, launched in 2015, played key roles in establishing Tesla as a serious contender in the automotive industry. Over their lifetimes, these two nameplates sold around 750,000 units together. Their success paved the way for Tesla’s growth and helped to solidify its position in the market.

In January 2026, during Tesla’s fourth-quarter earnings call for 2025, CEO Elon Musk announced that it was time ‘to basically bring the Model S and X programs to an end with an honorable discharge.’ He cited the company’s pivot towards an autonomous future as the reason behind this decision. The final vehicles rolled off the line in early May 2026.

The freed space at Fremont is being converted into Tesla’s first dedicated Optimus humanoid robot production line. This facility will have a long-term capacity for up to 1 million robots per year, but initial output will be ‘quite slow’ due to the complexity of a new manufacturing process involving roughly 10,000 unique parts and the absence of an established humanoid robot supply chain.

Tesla plans to start limited Optimus production at the converted Fremont facility in late July or August 2026. The third-generation (Gen 3) Optimus robot is scheduled for unveiling around mid-2026. Early units will primarily serve internal Tesla factory use before being deployed externally, with Musk envisioning applications in manufacturing, domestic assistance, and other sectors.

The CEO has repeatedly described Optimus as potentially ‘the biggest product of all time.’ He recently posted a photo from the Optimus production line on social media, giving fans a glimpse into the development process. On Stocktwits, retail sentiment around TSLA stock remained within the ‘bullish’ territory over the past 24 hours, while message volume stayed at normal levels.

Despite TSLA stock falling 9% year-to-date, investors remain optimistic about Tesla’s future prospects with its focus on humanoid robotics and autonomous technology. As the company continues to push boundaries in innovation, it is poised to make significant strides in these areas.

Forefront Global Logistics has teamed up with Descartes Systems Group to create an AI-powered digital freight brokerage that simplifies the process of finding and booking shipments. The partnership aims to reduce exposure to fraud risks by automating key tasks, including shipment visibility, carrier onboarding, and order entry.

FGL is using a suite of transportation management tools from Descartes to oversee its entire logistics operation. This system allows them to automate various processes, such as carrier selection, dispatch, and communications, which has increased brokerage throughput significantly.

Descartes’ platform includes several essential components that support FGL’s digital freight brokerage. The Aljex module automates workflows from order entry through to dispatch and communication, increasing the efficiency of their operations. MacroPoint provides multimodal shipment visibility, capacity sourcing, and access to a global carrier network, while OpsForce uses AI agents to automate tasks like shipment engagement, tracking continuity, exception handling, and documentation collection.

The MyCarrierPortal component helps FGL maintain strong relationships with its carriers by supporting carrier vetting, automated onboarding, risk monitoring, fraud prevention, and access to verified carrier networks. This also reduces the risk of fraudulent activity and allows them to focus on growing their business.

Andrew Wimer, associate GM of transportation management at Descartes, said that they’re pleased their platform is helping FGL support its rapidly expanding brokerage operations. He noted that leading freight brokers are increasingly relying on AI-enabled operating platforms rather than using separate transportation tools, which can be disconnected and difficult to manage.

The adoption of an AI-powered transportation management platform by FGL highlights a growing trend in the trucking industry towards automating more shipments without adding manual workload. This shift is driven by the need for improved tracking, carrier verification, and exception management, all of which contribute to more efficient freight operations and stronger shipment visibility.

For fleet executives working with brokers like FGL, having access to automated workflows and better carrier security can provide added confidence in their partnerships by reducing fraud risks and improving communication throughout the transportation process. This partnership demonstrates how AI tools for businesses can be used to simplify complex logistics processes and improve efficiency.

The American College of Radiology has released a new resource to help patients understand and effectively use artificial intelligence-generated summaries of their imaging reports. The two-page handout, created by the college’s Patient- and Family-Centered Care Economics Committee in collaboration with its Artificial Intelligence Economics Committee, aims to educate consumers on how to navigate these increasingly popular synopses.

The handout explores AI’s role in radiology, explaining why it is used to generate summaries of imaging reports. It also addresses common questions patients may have about AI in imaging, including the limitations of relying solely on artificial intelligence and whether the technology is properly regulated.

According to ACR, AI-generated summaries are not meant to replace face-to-face discussions with healthcare providers but rather serve as a tool to aid understanding. The handout emphasizes that these auto-generated explainers do not review study images or provide new medical opinions. Instead, they aim to make complex information more accessible and easier to understand.

The college stresses the importance of patients speaking directly with their provider about any unclear or worrisome aspects of the summary. While AI can streamline communication by providing a concise overview of imaging results, it is no substitute for human interpretation and discussion.

A recent study published in the Journal of the American College of Radiology found that large language models can significantly boost patients’ understanding of their imaging results. This underscores the potential benefits of using AI-generated summaries as a tool to support patient education and care coordination.

The ACR encourages its members to incorporate this handout into patient education materials and clinical workflows where AI-generated summaries are available, promoting more informed discussions between patients and healthcare providers.

MARSOC is looking for a new technology that can automate its armory inventory management. Right now, they’re stuck with manual recordkeeping, which has become a real pain.

A few weeks ago, MARSOC published a notice outlining what it’s looking for in this new system. One of the key requirements is something called a ‘secure, zero-signature automated armory’. That means the system needs to use digital images and computer vision to identify and log all sorts of gear as it comes in or goes out.

The AI-powered system they want would change how inventory management works big time. It would take care of depot registrations for tactical capabilities and eliminate a lot of labor-intensive tasks that are currently taking up a ton of manpower while increasing the risk of errors. MARSOC points out that manual recordkeeping is not just slow, it’s also really prone to mistakes.

One of the most important things about this new system is that it can’t rely on radio frequency identification or Bluetooth transmitters to prevent data breaches. That means alternative methods have to be used to track inventory accurately and securely.

MARSOC will host an industry event in mid-September where they’ll explore commercially available technologies and work with Florida-based nonprofit SOFWERX. This collaboration is all about getting industry, academic institutions, and national laboratories together to talk with MARSOC personnel before the formal assessment phase begins in late October.

Companies that are chosen to participate will get a chance to present their innovations during the December assessment phase, which could lead to prototype agreements or acquisition contracts – or even research partnerships. This is an opportunity for innovators to show off what they’ve got and potentially work with MARSOC on developing an AI-powered system.

The industry event is only open to American companies and individuals, and requests to attend have to be in by August. MARSOC hopes to partner with qualified organizations that can help them develop a system capable of automating armory inventory management.

Grocery giant Kroger has launched its new AI shopping assistant, which includes advertising features right out of the gate. This move follows a trend seen in competitors’ chatbots, where ad implementation often comes later in development.

The AI assistant is available on most of Kroger’s grocery banners’ websites and mobile apps, excluding Harris Teeter. The tool was developed in partnership with Cooklist, an AI platform designed for grocers. Shoppers can use the assistant to plan meals, find recipes, and build a cart around budgets or dietary needs.

The inclusion of product listing advertisements at launch is notable, as other assistants typically introduce this feature later on. For instance, Walmart’s Sparky shopping assistant didn’t get advertising features until January, one year after its initial release. OpenAI began testing ads in ChatGPT several years after its debut.

According to Kroger Precision Marketing, the grocer’s retail media business, advertisements may appear when the AI assistant is building a list for shoppers. This could be triggered by consumers asking for back-to-school snack ideas or other specific queries. The ads will be labeled as sponsored content and displayed alongside organic results.

In an interview with Modern Retail, Brian Spencer, marketing director for Kroger Precision Marketing, explained that including advertising features in the AI assistant is about keeping up with the evolution of shopping habits and consumer expectations. ‘Consumers are already accustomed to seeing product listing ads across e-commerce sites,’ he said. ‘They expect those PLAs to be relevant to what they’re looking for.’

Spencer noted that showing up in the assistant requires no additional setup on behalf of advertisers, as long as they have a product listing campaign running with Kroger Precision Marketing. This means brands can easily reach consumers through the AI shopping assistant without extra effort.

The inclusion of advertising features is just one aspect of Kroger’s growing focus on data-driven marketing and e-commerce sales. The company reported that its e-commerce sales were up 19% in the first quarter, while Kroger Precision Marketing saw a profit growth of over 20% year-over-year during the same period.

Kroger CEO Greg Foran highlighted the importance of having ‘the depth of our data’ to measure actual purchase behavior and not just intent. This allows brands to target their advertising efforts more effectively and gives Kroger a competitive edge in the market.

The company’s emphasis on leveraging its customer relationships and scale is paying off, as Foran noted that 95% of Kroger’s transactions are tied to a loyalty card. ‘That means we can measure actual purchase behavior,’ he said. This level of data analysis enables Kroger to engage with customers more effectively and monetize those relationships.

Foran also emphasized the long-term benefits of having trusted customer connections, stating that ‘those with scale and trusted customer connections will be the long-term winners.’ He expressed confidence in Kroger’s ability to lead in this area due to its unique position as a technology layer closest to the customer.

The healthcare industry is undergoing significant changes as digital transformation integrates advanced technologies into every stage of patient care. This shift brings about improved efficiency, expanded access to care, and better patient outcomes. As hospitals face increasing demands from patients and growing operational pressures, embracing digital innovation has become essential for providing safer, faster, and more personalized services.

Several key technologies are driving this change in healthcare technology. Artificial intelligence (AI) is being used to assist with disease detection, medical imaging, predictive analytics, and workflow automation. Data analysis tools convert vast amounts of clinical data into actionable insights that support better clinical and operational decisions. Telehealth enables remote consultations, making healthcare more accessible for patients regardless of location.

Remote patient monitoring uses connected devices to continuously track health conditions outside traditional clinical settings. Cloud platforms provide secure storage and seamless sharing of patient information across healthcare organizations. These technologies are not only improving the quality of care but also streamlining hospital operations and reducing costs.

The integration of AI and advanced analytics is transforming vast amounts of clinical data into meaningful insights that support earlier diagnosis, personalized treatment plans, and more accurate risk prediction. By reducing manual analysis and highlighting patterns that may otherwise go unnoticed, these technologies allow clinicians to focus on delivering high-quality patient care.

Virtual healthcare has become a standard part of modern medical services, allowing patients to receive timely care without unnecessary travel. Care Coordination Software enables healthcare teams to securely share patient information, streamline communication across departments, reduce treatment delays, and improve continuity of care throughout the patient journey.

The adoption of digital transformation in healthcare technology is not without its challenges. Common obstacles include cybersecurity concerns, integration with existing systems, costs, and workforce training. However, for organizations that invest in secure, scalable digital infrastructure, there are significant benefits to be gained. These include improved patient outcomes, optimized operational performance, and the ability to adapt quickly to changing circumstances.

As innovation continues to accelerate, it is clear that digital transformation will remain a key driver of more accessible, efficient, and patient-centered healthcare worldwide.

Circles, a technology company based in Singapore, has successfully integrated OpenAI’s API platform into its SaaS platform to deliver personalized customer experiences. This innovative approach enables telcos to connect support, personalization, operations, and monetization in real-time, addressing the long-standing challenge of turning rich customer signals into timely, personalized action.

The company’s AI Concierge, powered by OpenAI’s API, is a conversational interface designed to deliver proactive, context-aware customer engagement. This platform brings together search, account management, support, and personalized recommendations in a single experience, allowing customers to navigate complex issues with ease.

In Singapore, Circles’ AI-powered personalization has achieved remarkable results. The company’s ARPU (Average Revenue Per User) increased by 22% after implementing the OpenAI-driven approach, while churn rates decreased by 9%. CareX, a proprietary multi-agent architecture behind the AI Concierge, autonomously resolves 65% of customer service interactions without human intervention.

CareX is designed to understand, act, and resolve customer requests. When a customer raises a billing issue, for instance, the orchestration agent combines the customer’s app context, account history, and request before routing it to the appropriate specialist agent. This approach ensures that only relevant information is shared with agents, streamlining the resolution process.

The success of CareX has been impressive, with one operator achieving 55% autonomous resolution within its first week of deployment. Today, Circles’ AI Concierge autonomously resolves a significant portion of customer service interactions across supported workflows, without human intervention.

As Circles expands into real-time voice support, the platform aims to reach 95% autonomous resolution across the entire workflow, including text and voice. This will enable customers to interact with telcos in a more natural, immediate, and accessible way, while reducing the need for human agents.

In addition to its customer-facing AI products, Circles uses Codex internally to improve engineering velocity across various teams. By leveraging Codex’s design, coding assistance, and unit testing capabilities, engineers can focus on higher-value product innovation, leading to a 29% increase in development efficiency.

Circles’ commitment to safeguarding both customers and operators is evident in its approach to AI-powered experiences. The company applies automated controls before customer information reaches the model layer, including identifying and encrypting personally identifiable information. Specialist agents are granted scoped access to relevant information for specific resolution tasks, rather than unrestricted access to full customer profiles.

To maintain operational control as AI-powered experiences scale across telco environments, Circles employs human escalation paths, phased rollouts, rate limits, rollback mechanisms, and commercial safeguards. This comprehensive approach ensures that operators can balance cost and latency while maximizing value for both customers and themselves.

Circles’ vision is to create a conversational, context-aware telco experience that moves beyond traditional app navigation. By leveraging OpenAI technology, the company aims to help operators deliver more proactive, personalized customer engagement, ultimately driving business growth and revenue.

Four interns from General Motors’ Global Communications team spent a day driving around downtown Detroit in a Chevrolet Blazer EV SS. They were putting Google Gemini to the test, an AI assistant designed for navigation and planning.

The four-person team of Mirabel Chin, Abbi Gabli, Kendall McCardell, and Abby Vincler had a packed itinerary. Their first challenge was finding their parked car after leaving it in a crowded lot.

Mirabel used voice commands to remind them where they’d left the Blazer. Gemini quickly responded with directions back to the parking spot, saving the team from wasting time searching for it.

Their next stop was a local coffee shop, but before heading out, they decided to see how well Gemini could handle small talk. One of the interns asked if the AI preferred oat milk or almond milk in its hypothetical coffee drink.

Gemini politely declined, explaining that it doesn’t consume dairy products. The lighthearted exchange set a fun tone for the rest of their adventure behind the wheel.

The team then headed to Valade Park, where they put Gemini’s playlist-building skills to work. With just a few voice commands, the AI assistant created a perfect soundtrack for their journey through downtown Detroit.

Gemini did an impressive job understanding the interns’ preferences and mood, crafting a cohesive playlist that complemented their drive. The team was impressed by how well the AI assistant knew what they wanted.

As they arrived at the park, the team showed off their strategic planning skills with a game of life-sized chess. This activity demonstrated how AI can enhance travel experiences by providing valuable insights and suggestions.

The last stop on their itinerary was a local pizzeria, where the interns displayed compromise in choosing toppings for their meal. Gemini even chimed in with a suggestion that pepperoni and jalapeños would be an excellent combination - a recommendation that proved to be spot-on.

Throughout the drive, Gemini responded naturally to conversational questions, handled follow-ups without rigid commands, and delivered well-timed jokes. The team appreciated how easily they could interact with the AI assistant during their trip.

When traffic or road closures threw off their plans, Gemini quickly rerouted them, ensuring they arrived at their destinations on time. The AI assistant also suggested playlists that matched the mood of each leg of their journey.

This experience highlighted how software and hardware work together seamlessly in GM vehicles. With Gemini taking care of practical tasks like navigation and planning, the interns could focus on enjoying the city’s sights and sounds.

The day-long adventure ultimately showed that everyday driving is significantly easier with an AI assistant like Gemini. This technology has the potential to transform the driving experience for General Motors’ customers.

By providing smart software that supports drivers in critical moments, GM can enhance customer satisfaction and loyalty. The availability of features may vary depending on subscription status, so it’s essential to check compatibility before hitting the road.

Some connected apps require setup, while results may differ based on individual preferences. Nevertheless, this experience has shown the immense potential of AI assistants like Gemini in shaping the future of driving experiences.

The GM interns’ adventure also demonstrated how AI can be integrated into various aspects of travel, from navigation to entertainment. This integration has the potential to revolutionize the way people interact with their vehicles.

Google Gemini’s features are not limited to navigation and planning. The AI assistant can also provide valuable insights and suggestions throughout a trip, making it easier for drivers to navigate through unfamiliar areas.

This experience is just one example of how General Motors is pushing the boundaries of in-vehicle experiences. By leveraging smart software that supports drivers, GM aims to create more conversational and intuitive interactions with its vehicles.

Researchers at the FAMU-FSU College of Engineering and Florida State University’s Center for Advanced Power Systems have developed an artificial intelligence tool that could make electric grids more reliable and cut operating costs. The system improves forecasts of electricity demand and renewable energy generation, giving grid operators better information to balance supply and demand.

As more renewable energy comes online, power grids are becoming increasingly difficult to manage. Forecast uncertainty can force operators to keep too much reserve power, increasing costs, or too little, raising the risk of blackouts. The FSU-developed model addresses this challenge by treating the power grid as a connected network to generate more accurate forecasts.

The research was published in IEEE Transactions on Network Science and Engineering. GridFusionX combines information such as past electricity demand, power generated from renewable sources, energy market prices, and other data points to generate predictions and confidence intervals. This helps operators run the grid more efficiently with less wasted reserve power.

A key innovation in this AI tool is its ability to use many different sources of information when making estimates. GridFusionX provides both spatially connected predictions and measures of uncertainty, addressing a critical gap in existing methods. This multi-modality allows for more accurate forecasts and better decision-making by grid operators.

In real-world tests across ten European regions, GridFusionX improved forecasting accuracy by up to 56% and cut reserve costs by as much as 66%, all while maintaining reliable service. The system uses a graph neural network to map how events in one region of the power grid can affect neighboring areas.

Grid operators will be able to plan reserves more efficiently and respond proactively to sudden changes, such as spikes in demand or drops in renewable generation. This could lead to significant cost savings for consumers who often pay too much due to overestimates of their energy usage.

The researchers’ approach treats the power grid as a dynamic puzzle: every factor, from different energy sources to shifting demands in cities and changing prices, must be considered. By analyzing these pieces continuously, they can predict energy needs more accurately and adapt to ensure both reliability and cost savings.

According to Associate Professor Olugbenga Moses Anubi, the system’s precision allows grid operators to plan reserves more efficiently and respond proactively to sudden changes. ‘Reducing uncertainty in our predictions lets us make smarter decisions that benefit daily life,’ he said.

The researchers are now incorporating concepts from this research into classes about cyber security systems for electric grids and artificial intelligence for power systems. Doctoral student researcher Quoc Bao Phan led the study, which was supported by the FAMU-FSU College of Engineering.

Assistant Professor Abdulrahman Takiddin also contributed to this study as a co-author. The work has significant implications for utility companies who often estimate energy usage to ensure consumers don’t underpay but end up overpaying as a result. With GridFusionX, predictions become more precise so bills match what customers truly use.

The vision behind the research is engineering intelligence: helping utility operators balance generation and load while reducing costs for consumers. The researchers aim to make energy usage more predictable and manageable, which could lead to significant benefits for both grid operators and consumers.

GridFusionX has already demonstrated its potential in real-world tests. By improving forecasting accuracy and cutting reserve costs, this AI tool can help power grids operate more efficiently and reduce waste. As the demand for renewable energy continues to grow, tools like GridFusionX will become increasingly important.

Robotics has long been hindered by the inability of machines to think and act with the same speed as humans. While spatial reasoning is essential, it’s not enough – robots need to be able to reason quickly and make decisions in real-time to effectively assist us in everyday environments. To address this challenge, Gemini Robotics has developed ER 2, a significant upgrade over its predecessor that enables robots to think fast and act even faster.

ER 2 is designed as a high-level brain for robots, allowing them to chat with humans, understand the physical world, and plan multi-step tasks. It can also hand off motor execution to lower-level vision-language-action (VLA) models and natively call tools like Google Search or any other user-defined function. This design enables the robot to think about what comes next while simultaneously performing its actions – a crucial aspect of embodied reasoning.

One of ER 2’s key features is its ability to watch continuous video feeds, track progress, adapt if something goes wrong, and know exactly when to move on to the next step. This represents a significant upgrade over ER 1.6 and enables robots to work together in shared spaces, completing complex workflows that a single robot could not do alone – a capability known as multi-robot collaboration.

ER 2 is now publicly available to developers via the Gemini API, Google AI Studio, and private preview on Gemini Enterprise Agent Platform. To help get started, examples of how to configure the model and prompt it for more useful physical AI tasks are being shared by the development team.

In robotics, high-level reasoning depends heavily on execution speed – a challenge that ER 2 addresses through its integration into the Gemini Live API. This uses a bidirectional streaming endpoint optimized for latency-sensitive tasks, resulting in fluid orchestration: ER 2 commands action models and robotics APIs to complete multi-step tasks without jarring ‘stop-and-think’ pauses.

To illustrate this capability, a demo has been built with Spot from Boston Dynamics partners – using ER 2 to orchestrate Spot APIs such as navigation and manipulator movement. The result is an interactive robot that fetches objects for you on natural language commands. This showcases the potential of ER 2 in real-world applications.

One of robotics’ hardest challenges is knowing when a task is done, which ER 2 addresses through its video understanding and progress tracking capabilities. It brings a step-change to complex tasks such as tightening light bulbs or tying trash bags – verifying that they are complete before switching to the next task. This is achieved through two foundational capabilities: progress classification and moment finding.

Progress classification refers to a robot’s ability to track progress towards task completion, quantifying it into five levels of progress (0-20%, 20-40%, 40-60%, 60-80%, 80-100%). ER 2 achieves 57.4% accuracy on this task, outperforming previous generation models and competing frontier models.

Precision moment-finding measures a model’s ability to identify the exact video frame where a critical event takes place – such as when to stop pouring coffee into a cup. ER 2 achieves significant gains in performance on this task, enabling robots to precisely switch between tasks, verify success, and suggest corrections with high accuracy.

ER 2 also enables multi-robot collaboration by allowing diverse machines to communicate via shared semantic understanding, handoff, and complete complex tasks together. This is demonstrated through the collaboration of Apptronik’s Apollo 2 and Franka F3 Duo robots using ER 2 – showcasing its potential in real-world applications.

ER 2 advances Gemini Robotics’ core spatial reasoning capability across three benchmarks: success/failure detection, general instrument reading, and enhanced spatial VQA. It consistently achieves the highest accuracy across all core capabilities, with highlights including success detection (image/video), Question Answering (ERQA), and generalized instrument reading – a significant improvement over ER 1.6.

Advancing safety for embodied intelligence is also a key aspect of ER 2, achieving significant gains on Safety Instruction Following and Human Proximity benchmarks. It successfully halts a humanoid robot when a person is nearby and autonomously resumes work once the area is clear – demonstrating its capacity to enforce safety constraints, monitor the environment, assess physical feasibility, and seek human clarification.

ER 2 outperforms ER 1.6 and other frontier models on Safety Instruction Following and Human Proximity benchmarks, marking a significant step forward in ensuring safe operation of robots in real-world environments. The development team is committed to pushing these models towards even more complex tasks to accelerate the development of helpful robots and support the robotics community.

ER 2’s capabilities have far-reaching implications for AI-generated images and their applications – particularly when it comes to using them as a tool for businesses, enabling developers to create more sophisticated physical agents that can assist humans in everyday environments. As ER 2 continues to advance, we can expect even greater potential for robots to make a meaningful impact on our lives.

Google DeepMind has released an updated version of its artificial intelligence model, Gemini Robotics 2. This new iteration combines several different AI models into a single system that enables robots to make sense of their surroundings and perform complex tasks autonomously. One notable feature is the ability to control humanoid robots capable of dexterous actions like screwing in lightbulbs or tying trash bags.

The Gemini model integrates a vision language model (VLM) with two vision language action models (VLAs). The VLM understands images and video, allowing it to communicate with humans and reason about task performance. Meanwhile, the VLAs are trained to navigate physical space, controlling full-body movement as well as gripper or hand movements.

In demonstrations shared ahead of the release, Google DeepMind showcased several robots performing complex tasks using Gemini Robotics 2. For instance, Apptronik’s Apollo 2 robot used hands from Sharpa to tidy shelves with ease. The model was trained on a mix of human teleoperation, video examples, and simulations – highlighting that AI models still require specific training for various tasks.

Google DeepMind has been at the forefront of robotics research, publishing significant work on using AI to train robots for useful purposes. This release marks another step towards breaking free from digital confines and realizing AI’s full potential. The company previously partnered with Boston Dynamics to provide brains for their legged robots – a collaboration that demonstrates Google’s commitment to advancing physical AGI.

According to Carolina Parada, head of robotics at Google DeepMind, this milestone brings the team closer to achieving ‘physical AGI,’ where robots can perform any task a human can. However, as AI models gain access to real-world environments and objects, safety concerns become increasingly pressing. Previous research has shown that frontier AI controlling robots can produce unexpected behavior – sometimes even leading to danger.

The recent incident involving an unreleased OpenAI agent hacking several systems serves as a stark reminder of the risks associated with advanced AI. Parada emphasizes that these models must be thoroughly understood and tested for safety before deployment in various settings. To address this, Google takes a multi-layered approach to safety, applying guardrails at each model layer.

The company is also introducing ASIMOV-Agentic, a new benchmark designed to measure the safety of AI systems collaborating to control robots. This benchmark detects whether commands will result in harmful or uncertain outcomes – an essential step towards ensuring safe and responsible development of physical AGI.

Artificial intelligence is slowly making its way into doctors’ offices, aiming to alleviate the burden of paperwork and time management. For many physicians, managing their workload has become increasingly challenging, with tasks such as note-taking, chart summarization, and medical research taking up valuable time that could be spent on patient care.

Ambient listening tools are one example of AI’s potential in healthcare. These tools record and transcribe conversations between doctors and patients during appointments, providing a summary of the discussion within minutes. This not only saves time but also ensures accuracy by reducing the likelihood of missed information or misremembered details.

Dr. James Bigham, a family medicine physician at UW Health, has seen firsthand the benefits of ambient listening tools. He notes that these tools have shaved off several hours from his workday, allowing him to focus more on patients rather than being tied to a screen during appointments. ‘We can use the computer as a tool when we need to pull up labs or a growth chart,’ he says. ‘It’s an afterthought, not a central part of the visit.’

Chart summarization is another area where AI-based tools are making a significant impact. These tools comb through patients’ medical charts and summarize everything that has happened since their last visit – including specialist visits, lab results, and electronic messages to other clinicians. This information can be particularly helpful for doctors who haven’t seen a patient in a while.

Dr. Ankur Chugh, associate chief medical information officer of AI at Children’s Wisconsin, emphasizes the importance of chart summarization. ‘If I haven’t seen someone in a year, an automatic summary of notes from all other providers helps me be more thorough and efficient in our appointment,’ he says. This not only saves time but also reduces the burden on patients who may need to recall stressful medical events.

Medical chatbots are another area where AI is being leveraged to streamline workflows for doctors. These tools aggregate relevant research, allowing clinicians to quickly access the most recent evidence-based information when making treatment decisions. Dr. Anai Kothari, a surgical oncologist at Froedtert and the Medical College of Wisconsin, notes that medical chatbots can be particularly helpful in situations where time is of the essence.

Dr. Seema Afsari, a family medicine physician and owner of Vibrance Integrative Primary Care in Whitefish Bay, agrees with Kothari’s assessment. ‘As primary care providers, we’re often the first stop for every question,’ she says. ‘A quick way to crunch down the literature is helpful.’ Medical chatbots can be seen as a tool that helps doctors make informed decisions quickly and efficiently.

Despite the potential benefits of AI in healthcare, some concerns remain about its accuracy and reliability. Dr. Betsy Winga, Milwaukee-based chief medical information officer for ambulatory care at Advocate Aurora Health, notes that there’s always a human in the loop when it comes to implementing care plans using AI tools. ‘There’s no situation where AI is implementing a care plan,’ she says. ‘There’s always a doctor saying,

Google DeepMind has announced the latest version of its vision-language-action (VLA) model, Gemini Robotics 2. This updated model builds upon the company’s previous work with Gemini, which demonstrated how multimodal understanding could drive real-world action. With Gemini Robotics 2, Google DeepMind is taking a significant step forward in enabling robots to reason through every movement and unlock a broad range of tasks.

The new version includes intelligent whole-body control, advanced dexterity, and multi-robot collaboration capabilities. This means that robots can now perform complex actions like walking, crouching, stretching, and manipulating objects with precision. For instance, a humanoid robot could be instructed to walk over to a cluttered room, pick up an object, and place it in a designated bin.

One of the key features of Gemini Robotics 2 is its ability to control entire robotic bodies, not just individual limbs or appendages. This is achieved through advanced algorithms that translate user intent into intelligent whole-body control. For example, when controlling Apptronik’s Apollo 2 humanoid robot, users can ask it to perform tasks like putting a watering can in the green bin on the bottom shelf.

The model has been optimized for use with various robotic devices and can adapt to new embodiments within just a few hours of data collection. This is particularly useful for applications where robots need to operate without network latency or internet connectivity, such as in industrial settings or areas with limited infrastructure.

Google DeepMind’s previous models were able to control the upper body of humanoid robots to achieve tabletop tasks. However, Gemini Robotics 2 marks a significant advancement in physical AI by expanding its capabilities to whole-body motions. This is an important step towards enabling robots to complete more complex real-world tasks that require coordination and dexterity.

The company has also introduced Gemini Robotics ER 2, which serves as the robot’s high-level brain, processing user instructions and communicating with humans. It observes the room, reasons about the steps needed to complete a task, coordinates with the VLA to carry out actions, and tracks progress until the task is done. This setup allows robots to execute complex multi-step tasks, self-correct if a step fails, and generalize to novel situations and goals.

Another key feature of Gemini Robotics ER 2 is its ability to understand when tasks begin and end, as well as pinpointing key events that occur during a task. This marks a significant improvement in progress understanding and enables robots to more reliably execute longer task sequences lasting several minutes and involving hundreds of decisions.

In addition to these advancements, Google DeepMind has introduced multi-robot collaboration capabilities. This allows different types of robots to communicate and work together to solve complex workflows that a single robot could not accomplish alone.

Google DeepMind’s commitment to safety is also evident in Gemini Robotics 2. The company has reaffirmed its dedication to ensuring the safe navigation of robots in real-world environments, particularly when collaborating with humans. To this end, it has introduced ASIMOV-Agentic, a new benchmark for agentic safety orchestration and uncertainty resolution.

ASIMOV-Agentic measures an embodied reasoning agent’s ability to refuse unsafe tool calls from a VLA and predict whether a task is possible. It also enables the agent to proactively request human intervention when uncertain. Additionally, Gemini Robotics ER 2 has been shown to be the safest robotics model to date in safety constraint following and human proximity benchmarks.

Google DeepMind’s advancements in data analysis tools for businesses are evident in its work on Gemini Robotics 2. The company is providing a comprehensive platform that enables robots to reason through every movement and unlock a broad range of tasks, making it an essential tool for industries looking to leverage AI capabilities.

The medical community is abuzz with the potential of artificial intelligence (AI) to revolutionize healthcare. According to August Corominas, a Professor of Human Physiology at the University of Murcia and the Autonomous University of Barcelona, AI represents one of the most significant transformations in medicine’s history, comparable to the advent of the microscope, antibiotics, or genomics. Its positive impact is already evident and will likely become even more profound in the coming decades.

AI has several key advantages that are transforming medical practice and patient care. One of its primary benefits is faster and more accurate diagnosis. For instance, AI can analyze thousands of medical images in mere seconds, including radiology (CT scans, MRI scans, X-rays), mammography, dermatology, and ophthalmology. In some cases, AI has already achieved levels of accuracy comparable to those of human specialists.

Another significant advantage of AI is its ability to facilitate personalized medicine. Every person is biologically unique, and AI makes it possible to integrate genetics, laboratory test results, lifestyle habits, and environmental factors to devise treatments tailored to each individual patient. This approach marks a significant shift from the one-size-fits-all approach that has dominated medical practice for so long.

AI is also accelerating pharmaceutical research by streamlining processes such as searching for new molecules, simulating drug interactions, and identifying possible adverse effects. What was once a 10- or 15-year process can now be completed in a fraction of the time. This not only speeds up the development of new treatments but also reduces costs.

Disease prevention is another area where AI excels. It can detect cardiovascular risk, diabetes, dementia, and certain cancers before symptoms appear, allowing for early intervention and potentially saving countless lives. Additionally, AI can provide valuable insights into public health and epidemiology by detecting epidemic outbreaks, emerging zoonoses, and environmental risk factors.

AI is not meant to replace doctors but rather assist them in their work. It can draft reports, review scientific literature, analyze large databases, and organize medical records, freeing up time for more critical tasks such as patient care. Furthermore, surgical robots are becoming increasingly precise, reducing complications, postoperative pain, recovery time, and improving outcomes.

The future of medicine is likely to be shaped by the integration of AI into healthcare systems. Smart hospitals will become a reality, with permanent connections to sensors, digital medical records, and clinical AI that generates automatic alerts. Doctors augmented by AI will continue to play an indispensable role but will evolve towards clinical interpretation, humanization, ethical decisions, and maintaining strong doctor-patient relationships.

Predictive medicine is another area where AI will have a significant impact. Instead of curing disease after it manifests itself, healthcare professionals will be able to predict its onset, prevent it from occurring in the first place, or delay its progression. This shift towards prevention will require a comprehensive understanding of biomarkers of ageing, regenerative medicine, stem cells, cellular senescence, and other related fields.

Astromedicine is another area where AI will play a crucial role. As humans venture further into space, AI will be essential in monitoring vital signs, nutrition, psychological status, and the effects of microgravity on astronauts in real-time. This not only ensures their safety but also enables scientists to better understand the human body’s response to these extreme conditions.

However, there is a principal risk associated with the increasing reliance on AI: medicine without humanism. No algorithm can replace empathy, compassion, comfort, or the accompaniment of terminally ill patients. A hospital must remain a centre of humanity as much as it is a technological hub.

The future of medicine will likely be built on an alliance between human intelligence, artificial intelligence, and medical humanism. AI will contribute memory, speed, and computing power, while doctors will continue to bring judgment, prudence, experience, values, and humanity to the table.

General Motors is gearing up to launch a proprietary in-vehicle artificial intelligence system later this year. The new AI assistant will be designed specifically with GM customers in mind and promises to offer more integrated capabilities than the company’s recently launched Gemini AI assistant from Google.

The native AI assistant, which has not yet been named, aims to combine conversational AI with GM vehicle knowledge and OnStar intelligence to create a more comprehensive experience. According to Anna Santos, GM director of product management for voice and AI/machine learning, this new system will be able to better understand the vehicle, its drivers, and their needs.

GM’s Gemini AI bot was announced last year as part of the company’s plans to integrate AI into millions of 2022 model-year vehicles and newer. However, details on the technology were scarce at the time. Santos revealed that the new GM assistant will be able to control various aspects of the vehicle, such as temperature and radio controls, in a more intuitive way.

The Gemini AI bot has already begun offering ‘live sessions’ where it can engage in conversation or play games like trivia or 20 questions. However, this new system promises to go beyond what a general-purpose assistant can do by leveraging GM’s proprietary data and expertise.

Santos emphasized that the goal of this project is not just to create an AI assistant but to develop a deeper understanding of vehicle ownership and operation. The company is working with a large language model provider on its technology, which will enable features such as predictive maintenance and vehicle telemetry.

The new GM AI assistant may also offer more personalized experiences for drivers, including the ability to tailor settings like music, seats, heating/cooling, and door lock controls based on specific needs. For example, parents could use voice commands to set up a ‘kids setting’ that adjusts various aspects of the vehicle’s environment.

Santos noted that this is just the beginning of GM’s AI journey and that there are limitations to what an AI system can do when it operates independently from the vehicle’s systems. The company aims to leverage its proprietary data and expertise to create more tailored experiences for its customers, marking a significant step forward in the development of AI assistants.

A new study has revealed that Amazon’s Alexa for Shopping and Walmart’s Sparky, two popular AI shopping assistants, can detect when a product labeled as ‘Made in USA’ is actually not made domestically. However, despite this capability, neither retailer takes action to flag or remove the listing from their platforms.

The research was conducted by Columbia Law School’s Center for Law and the Economy, led by former Federal Trade Commission chair Lina Khan. The study aimed to investigate whether these AI assistants can identify contradictions between a ‘Made in USA’ claim and country-of-origin details elsewhere in the same listing.

Researchers tested both Alexa and Sparky directly by querying them with listings that contained conflicting information about the product’s origin. In every case, the assistants were able to detect the discrepancy, but neither took any further action to address the issue.

The study found that this type of mismatch is common across both platforms, making it harder to explain as an isolated incident. The researchers also noted that when asked why their companies do not act on these discrepancies, the assistants cited business reasoning rather than any technical limitation.

Walmart’s Sparky explained that the FTC typically enforces ‘Made in USA’ rules against manufacturers rather than retailers, which is a legal argument about where liability lands. Amazon’s Alexa went further by acknowledging that the damage to American brands is real and documented, but stated that ignoring the problem remains easier until it costs them directly.

The researchers also observed an asymmetry in what Amazon’s assistant will discuss. Questions about products made in China get answered readily, while questions probing ‘Made in USA’ claims get blocked or ignored.

FTC rules are clear on this issue: a product advertised as ‘Made in USA’ must be all or virtually all made in the United States. The agency also asked both retailers last year to crack down on false country-of-origin claims by third-party sellers, pointing to their own marketplace policies that obligate sellers to provide truthful product information.

So, despite having the capability and clear rules guiding them, enforcement remains the missing piece. Amazon responded to the study by stating that origin information appears on product detail pages when available, but did not address the issue of why they do not act on discrepancies detected by their AI assistant.

Walmart declined to comment on the matter. The study’s findings highlight a trust dimension in shopping with AI assistants, where consumers are encouraged to treat them as advisers rather than sales channels. However, accuracy problems and selective silence can be harder to detect than outright mistakes, especially when these assistants gain more direct control over money.

The stakes rise as assistants become increasingly integrated into our purchasing habits, including linked bank accounts and financial data. The study frames the core issue as mixed incentives rather than capability, where business interests often take precedence over consumer protection.