Google Advances World Models While OpenAI Improves ChatGPT Features

Instacart recently halted its AI-powered pricing tests following an investigation by Consumer Reports, which revealed customers were seeing different prices for the same items. The study indicated that price variations, ranging from $0.07 to $2.56, could potentially cost families an additional $1,200 annually. This practice drew scrutiny from the Federal Trade Commission and several senators, including Amy Klobuchar. Instacart confirmed that all item price tests would cease immediately, ensuring uniform pricing for all customers, and denied using dynamic or surveillance pricing. The company also recently settled an FTC lawsuit for $240 million concerning delivery fees. In the broader AI landscape, Google DeepMind co-founder Shane Legg maintains his long-standing prediction of a 50% chance for Artificial General Intelligence (AGI) to arrive by 2028, envisioning a potential "golden age." Demis Hassabis, who leads Google's AI efforts, is actively seeking the next major AI application beyond chatbots, focusing on "world models" similar to those used in AlphaGo. Google's latest model, Gemini 3, shows strong performance, and Hassabis envisions future smart glasses that can understand and predict the physical world. AI is also transforming various sectors, including banking, where it enhances efficiency without causing significant job losses, though hiring growth may slow. A survey found 88% of finance leaders believe AI will be the most transformative technology in the next two years, despite a perceived skills gap in generative AI. In space engineering, Physics AI, championed by companies like Luminary Cloud, is revolutionizing design by simulating complex aerodynamics in seconds, accelerating the development of rockets and hypersonic systems. Meanwhile, OpenAI's ChatGPT has introduced a personalized year-end review feature called "Your Year with ChatGPT," available to users in select English-speaking markets. The legal framework for AI training and copyright is evolving, with several key court decisions expected in 2025. In `Bartz v. Anthropic`, a judge deemed AI model training "spectacularly transformative" for fair use, though using pirated books was not. Anthropic later settled for up to $1.5 billion. `Kadrey v. Meta` also found AI training transformative but warned of market harm. However, `Thomson Reuters v. ROSS Intelligence` ruled against ROSS for using copyrighted Westlaw content to train a competing AI tool, citing market harm. Additionally, the proliferation of AI-generated fake online reviews is decreasing the reliability of product feedback, necessitating consumer vigilance and the use of tools like Fakespot. Ontario is rapidly becoming a significant hub for AI investment, attracting companies like Cohere, which develops enterprise AI models, and Waabi, focused on self-driving trucks. Pharmaceutical giants AstraZeneca and Roche are also investing in AI for R&D in the region. Ontario added over 17,000 AI jobs and saw a 312% increase in new AI companies in the past year, fueled by its strong academic foundation and a large pool of AI and STEM graduates. To maintain global leadership in AI, Canada recognizes the critical need for increased investment in mathematics, which forms the core of AI and is essential for addressing issues of fairness, reliability, and sustainability.

Key Takeaways

• Instacart ceased its AI-powered pricing tests after a Consumer Reports investigation revealed customers saw varying prices, potentially costing families $1,200 extra annually, leading to FTC and senatorial scrutiny.
• Google DeepMind co-founder Shane Legg predicts a 50% chance of Artificial General Intelligence (AGI) by 2028, foreseeing a potential "golden age."
• Demis Hassabis, head of Google's AI efforts, is focused on developing "world models" to power future AI applications like smart glasses, moving beyond current chatbots.
• AI is transforming the banking sector by increasing efficiency without major job losses, though a skills gap in generative AI persists among finance leaders.
• Physics AI, utilized by companies like Luminary Cloud, is revolutionizing space engineering design by enabling rapid simulation of complex aerodynamics, accelerating development.
• OpenAI's ChatGPT launched "Your Year with ChatGPT," a personalized year-end review feature available to users in select English-speaking markets.
• Copyright law is actively addressing AI training; while some courts find it transformative, cases like `Thomson Reuters v. ROSS Intelligence` have ruled against using copyrighted content to train competing AI tools.
• AI is facilitating the mass production of fake online reviews, making it crucial for consumers to use tools like Fakespot and consult trusted sources for product assessments.
• Ontario is emerging as a major AI investment hub, attracting companies like Cohere and Waabi, and added over 17,000 AI jobs in the past year.
• Canada needs to increase investment in mathematics to maintain its leadership in AI, as a deeper understanding of its mathematical foundations is crucial for addressing issues of fairness, reliability, and sustainability.

Instacart stops AI price tests after consumer report

Instacart has ended its AI price experiment after a Consumer Reports investigation. The company was testing different prices for the same items using Eversight AI technology. Consumer Reports found that shoppers saw price variations from $0.07 to $2.56, potentially costing families $1,200 extra per year. Instacart also showed different "original" prices for discounted items. Instacart stated that all item price tests will conclude immediately and denied using dynamic or surveillance pricing.

Instacart stops AI pricing tests after public outcry

Instacart ended its AI-powered pricing tests after a Consumer Reports study revealed customers saw different prices for the same items. The study with 1,500 participants found 74% saw varying prices, which could cost a family of four an extra $1,200 annually. The Federal Trade Commission and several senators, including Amy Klobuchar, began investigating these practices. Instacart announced that all item price tests would stop immediately, ensuring all customers now see the same prices. The company also recently settled an FTC lawsuit for $240 million over delivery fees.

ChatGPT unveils "Your Year" review feature

ChatGPT, owned by OpenAI, launched a new year-end review feature called "Your Year with ChatGPT." This feature is similar to Spotify Wrapped and is available in select English-speaking markets like the US, Canada, and the UK. Users with free, Plus, or Pro plans who have chat history enabled can access it. "Your Year with ChatGPT" offers personalized "awards," a poem, and an image based on user activity. It is available on web, iOS, and Android apps, and users can also ask ChatGPT directly to trigger it.

Google DeepMind co-founder predicts AGI by 2028

Shane Legg, a co-founder of Google DeepMind, has consistently predicted the arrival of Artificial General Intelligence, or AGI. Since 2009, he has stated there is a 50% chance AGI will exist by 2028. Legg believes AGI could lead to a "golden age" with dramatically increased production of many things. However, he stresses that this potential prosperity depends on carefully translating machine capabilities into benefits for individuals and society. His prediction remains unchanged despite shifts in the AI industry.

AI changes banking but jobs remain stable

Experts say AI is transforming the banking sector without causing a major job loss. While banks are using AI for efficiency, they are not yet making deep workforce cuts, though hiring growth may slow for years. A survey found 88% of finance leaders believe AI will be the most transformative technology in the next two years. However, only a small percentage feel prepared for this shift, and over half see a skills gap in generative AI. Leaders like Tom Hood from AICPA & CIMA emphasize investing in talent and technology to turn disruption into a competitive advantage.

Physics AI revolutionizes space engineering design

Physics AI is rapidly changing space engineering, according to Juan Alonso, CTO of Luminary Cloud and a Stanford professor. Advances in computational fluid dynamics and physics AI allow designers to simulate complex aerodynamics in seconds. This greatly speeds up the design and testing of rockets, aircraft, and hypersonic systems. Luminary Cloud creates physics AI models that are faster and as accurate as older methods. This technology aims to accelerate design and certification, enabling quicker deployment of advanced systems for commercial and defense needs.

Copyright and AI training face legal challenges

Three important court decisions in 2025 show how copyright law is dealing with AI training. In Bartz v. Anthropic, a judge called AI model training "spectacularly transformative" for fair use, but ruled that using pirated books was not fair. Anthropic later settled for up to $1.5 billion. In Kadrey v. Meta, a judge also found AI training transformative but warned about potential harm to the market for original works. However, in Thomson Reuters v. ROSS Intelligence, the court ruled against ROSS for using copyrighted Westlaw content to train a competing AI tool, finding it harmed the market.

Check trusted sources for real online reviews

The reliability of online reviews is decreasing as AI makes it easier to create fake ones. While fraudulent reviews have existed for years, AI now allows for their mass production without human effort. Tools like Fakespot and ReviewMeta can help identify suspicious reviews. Experts advise shoppers to also consult trusted sources like Consumer Reports and CNet for unbiased product assessments. Consumers must remain watchful and careful when looking at online feedback to avoid misleading information.

Google AI leader Demis Hassabis seeks next big app

Demis Hassabis, co-founder of Google DeepMind and head of Google's AI efforts, is working to create the next major AI application. He believes "world models," like those used in AlphaGo, could power new AI technologies beyond chatbots. Google merged its AI units under Hassabis in 2023, and their latest model, Gemini 3, has shown strong performance. Hassabis envisions smart glasses that understand and predict the physical world, a significant leap from current devices. Google needs to find ways to make money from AI beyond advertising, and Hassabis's success could determine the company's future in the AI race.

Ontario attracts AI investment with talent and stability

Ontario is becoming a major hub for AI investment, drawing companies with its strong talent pool and stable environment. The Waterloo-Toronto-Ottawa corridor is seeing growth in AI, with companies like Cohere developing enterprise AI models and Waabi advancing self-driving trucks. Pharmaceutical giants AstraZeneca and Roche are also investing in AI for research and development. In the past year, Ontario added over 17,000 AI jobs and saw a 312% increase in new AI companies. This growth is fueled by the region's strong academic foundation, including Nobel laureate Geoffrey Hinton's work, and its large number of AI and STEM graduates.

Canada needs math investment to lead AI

To become a global leader in AI, Canada must invest more in mathematics, which forms the core of artificial intelligence. Many aspects of AI remain poorly understood, and progress relies on a deeper grasp of its underlying mathematical principles. While AI is widely used in daily life, it faces issues with fairness, reliability, and sustainability due to biases and high costs. Improving AI means understanding its complex mathematical "engine" rather than just building bigger computers. Canada can achieve this by funding science that makes AI systems predictable and efficient, leveraging its existing mathematical and AI institutes.

Sources

• Instacart Ends AI Price Experiment After Consumer Reports Investigation
• Instacart Ends AI Pricing Tests Showing Different Prices for the Same Items
• ChatGPT launches a year-end review like Spotify Wrapped
• Google Deepmind Co-founder Shane Legg Has Been Predicting AGI By 2028 Since 2009 — And He Still Sticks With It
• AI is reshaping banking—but not causing a jobs wipeout
• How physics AI is transforming the future of space engineering
• Copyright and AI Collide: Three Key Decisions on AI Training and Copyrighted Content from 2025
• Consult trusted sources to make sure online reviews are legit
• Google’s Chess Master Is Working on AI's Killer App
• AI attraction beyond the hyperscaler hype
• How can Canada become a global AI powerhouse? By investing in mathematics