Reddit is testing Reddit Reviews, AI-generated product pages built from discussions in its communities. The initial US rollout covers about 10,000 products across Electronics, Health and Beauty, Home and Garden, and Fashion and Apparel, on iOS, Android and the web. Pages summarize discussions, show positive, mixed or negative sentiment and common likes and dislikes, and link to original comments and related products. Users can discover pages through searches or product names in comments; Reddit says brands cannot pay for placement, though catalog data may supply prices and purchase links. AI Search can also answer product questions without a dedicated page. Reddit plans more products and categories through 2027 and expansion beyond the US.
Schools and universities are taking different approaches to AI use. Orange County Public Schools requires students to get an instructor’s permission and use district-approved platforms at school; the rules do not extend beyond school, and guardians are encouraged to ask students to explain their work. Maya Israel, involved in creating the Florida AI Education Task Force, says research has not established AI’s effects on critical thinking or relationships. At USC, a new minor jointly launched by Marshall and Dornsife combines sports, analytics and AI. Its courses include practical work with sports teams or technology companies and using AI to study trends and athlete performance.
AI’s potential in research comes with a need for verification. An essay in the European Journal of Epidemiology says large language models can help epidemiologists compare explanations, identify conflicting assumptions and generate alternatives, but even capable models can produce persuasive errors. Researchers should test AI-generated ideas against real-world evidence, formal reasoning, computation and reproducible analysis. A separate analysis argues that AI’s long-term value should not be judged by technology stock swings: Stanford estimates the cost of running models at a given benchmark performance fell more than 280-fold from November 2022 to October 2024. Its predictions include more AI colleagues, greater pressure on companies to demonstrate returns, AI-assisted scientific hypotheses, and more government investment in lifelong learning and social participation. National AI infrastructure spending could exceed $1 trillion annually by 2029, according to IDC.
OpenAI says its tools can identify software vulnerabilities, test fixes and prepare code changes for human review. The article reports that GPT6 Cyber reproduced 81 previously unknown high-severity Firefox flaws and raised the severity of 30 known issues; a Linux pipeline validated patches for almost 1,000 local privilege escalations. GPT6 Cyber scored 52% on Exploit Gym, while a newer model scored about 80% on PatchBench. Yet fixing one PHP flaw that had existed for 13 years took two months, underscoring the workload involved. OpenAI also cites Patch the Planet, a project with Trail of Bits, Calif and HackerOne, and a billion-dollar Daybreak defense fund for defenders.
AI-related spending and workplace monitoring are also drawing scrutiny. A RegTech Analyst report says 300 financial institutions expect to spend a combined $703.7 million on AI for risk and compliance in 2026—about 30% of their reported $2.37 billion in RegTech spending. Average expected AI spend ranges from $80,000 among firms with 50–249 employees to $11 million for those with at least 20,000. The report cautions that spending does not prove improved results and urges firms to measure outcomes such as faster investigations and less manual work. In New Zealand, hundreds of The Warehouse workers protested in Auckland and Christchurch over movement monitoring and fears of job replacement. The union says the retailer recently used drones for stocktaking, a packing robot and AI monitoring; some workers said tracking toilet breaks made them hold on.
In the UK, London- and Bristol-based Fractile is developing next-generation AI chips with backing from the Government’s Sovereign AI Venture Fund and the British Business Bank. The funding supports hardware intended for areas including healthcare, finance and transportation, with the project expected to finish within two years. Fractile is working with the government on developing the technology in a way that benefits society.
Key Takeaways
- Reddit’s US Reddit Reviews test covers about 10,000 products and links AI summaries and sentiment to the source community discussions; Reddit says placement is not paid.
- Orange County Public Schools requires instructor permission and district-approved platforms for student AI use at school; USC has launched a minor combining sports performance, analytics and AI.
- An epidemiology essay says AI can broaden research ideas but its outputs require checks against evidence, formal reasoning, computation and reproducible analysis.
- Stanford estimates the cost of running AI models at a fixed benchmark performance fell more than 280-fold from November 2022 to October 2024; IDC says annual national AI infrastructure spending could exceed $1 trillion by 2029.
- OpenAI reports GPT6 Cyber reproduced 81 previously unknown high-severity Firefox flaws and a Linux pipeline validated patches for almost 1,000 local privilege escalations; the article also cites Patch the Planet and a billion-dollar Daybreak defense fund.
- Financial institutions expect to spend $703.7 million on AI for risk and compliance in 2026, within $2.37 billion in RegTech spending; average AI spend ranges up to $11 million at the largest firms.
- Hundreds of The Warehouse workers protested AI monitoring and automation in New Zealand; reported technologies include stocktaking drones, a packing robot and worker monitoring.
- UK-backed Fractile is developing next-generation AI chips for potential healthcare, finance and transportation uses, with the project expected to be completed within two years.
Reddit tests AI product reviews built from community discussions
Reddit is testing Reddit Reviews, which uses AI to organize product discussions into review pages. The first US rollout will cover about 10,000 products in Electronics, Health and Beauty, Home and Garden, and Fashion and Apparel, reaching users on iOS, Android and the web over the next few weeks. Pages include a summary, a positive, mixed or negative sentiment breakdown, common likes and dislikes, related products, and links to original comments. Brands cannot pay to appear, though brand and retailer catalogs can provide prices and purchase links. Reddit plans to add products and categories through 2027 and expand beyond the US.
Reddit turns community discussions into AI product review pages
Reddit Reviews organizes existing subreddit discussions into product pages with sentiment, pros and cons, and comparisons. Users can find a page through product searches or highlighted product names in comments, and AI Search can answer questions about products without a dedicated review page. The rollout covers roughly 10,000 products in Electronics, Health and Beauty, Home and Garden, and Fashion and Apparel, with more categories and expansion beyond the US planned for 2027. Each page links to source discussions, while Reddit says the initial recommendations come from community conversations rather than paid placement.
USC launches minor combining sports performance and AI analytics
The USC Marshall School of Business and USC Dornsife College of Letters, Arts, and Sciences launched the Human Performance and AI in Sports Analytics minor this fall. Led by Lorena Martin, the program combines study of the sports industry, analytics, AI, and soft skills. Required course DSO 484 gives students practical experience with a professional, college, or intramural sports team, or a sports technology company. DSO 478 teaches students to use AI to study sports trends and methods for improving athlete performance. Martin says students need to understand both data and the human side of sports to use analysis well.
Central Florida schools require permission for student AI use
Orange County Public Schools introduced new AI rules for the current academic year. Students must get permission from an instructor before using AI and can only use district-approved platforms at school. The rules do not apply after students leave the school environment, so guardians are encouraged to ask children to explain their homework and show they understand it. Maya Israel, who helped create the Florida AI Education Task Force, says research has not yet established how AI affects students’ critical thinking or relationships. Some students say AI helps them, including a high school student using it to learn about brain function and neurological therapies.
Five predictions for how AI could reshape work and society
The article argues that AI’s long-term value should be judged separately from swings in technology stock prices. Stanford University estimates that the cost of running AI models at a given benchmark performance fell more than 280-fold between November 2022 and October 2024. The author predicts that AI colleagues will become common, companies will face more pressure to prove AI’s value, and AI will help produce scientific discoveries by generating hypotheses. The article also predicts that national AI infrastructure will become strategically important, with annual spending potentially exceeding $1 trillion by 2029, according to the International Data Corporation. Finally, it says governments may need to invest more in lifelong learning, human purpose, and social participation as AI changes work.
AI could widen epidemiology research but needs human checks
An essay in the European Journal of Epidemiology by Ulrich Mansmann explores how artificial intelligence could help epidemiologists consider more explanations and evidence. Large language models can compare competing ideas, find clashing assumptions, and suggest alternatives across different fields. But research cited in the essay found that larger and more capable models can still give convincing but incorrect answers that people may not catch. The essay says researchers must check AI-generated ideas against real-world evidence, formal reasoning, computation, and reproducible analysis.
OpenAI aims to find and fix software flaws faster with AI
OpenAI says its AI tools can find software flaws, test fixes, and prepare code changes for human review. GPT6 Cyber reproduced 81 previously unknown high severity Firefox flaws and raised the severity of 30 known issues, while a Linux pipeline validated patches for almost 1,000 local privilege escalations. OpenAI also reported that GPT6 Cyber scored 52% on Exploit Gym and that a newer model scored about 80% on PatchBench. The article notes that fixing a PHP flaw that had existed for 13 years took two months of work, showing the challenge of handling many flaws at once. OpenAI points to Patch the Planet, a project with Trail of Bits, Calif, and HackerOne, and a billion dollar Daybreak defense fund for defenders.
Warehouse workers protest AI monitoring in New Zealand
Hundreds of workers at retailer The Warehouse protested in Auckland and Christchurch on Thursday over technology that monitors their movements and may take over their jobs. The New Zealand Herald reported the protests, and Workers First Union told Radio New Zealand that the company had recently used drones for stocktaking, a packing robot, and AI to monitor workers. Some workers said AI tracking of toilet breaks forced them to hold on.
Financial firms spend more on AI, but returns remain unclear
A RegTech Analyst report found that 300 financial institutions expect to spend a combined $703.7 million on AI for risk and compliance in 2026. That is about 30% of the $2.37 billion in RegTech spending reported by a similarly sized group of institutions. Expected AI spending varies widely, from an average of $80,000 for firms with 50 to 249 employees to $11 million for those with at least 20,000 employees. Firms with 10,000 to 19,999 employees allocate the largest share of RegTech spending to AI, at 43.89%, while the largest firms allocate 25.89%. The report says spending alone does not show whether AI improves operations, and firms need to measure results such as faster investigations and reduced manual work.
UK funding backs Fractile’s next generation AI chips
Fractile, a startup based across London and Bristol, is developing advanced chips with funding from the UK Government’s Sovereign AI Venture Fund and the British Business Bank. The funding will support work on the next generation of AI hardware for uses including healthcare, finance, and transportation. The project is expected to be completed within the next two years. Fractile is working with the UK Government on developing the technology in a way that benefits society.
Sources
- Reddit Is Testing AI Review Pages Built From Its Threads
- Reddit Launches AI-Powered Product Review Pages Built From Community Discussions
- Human Performance and AI in Sports Analytics Minor Combines Practical Skills with Data Analysis
- Students face new AI rules at Central Florida schools
- Bubble or not, here are five predictions for what AI will really deliver
- Artificial Intelligence Expands Epidemiological Reasoning but Demands Rigorous Human Evaluation
- OpenAI wants AI to patch as fast as it finds
- New Zealand workers say AI tracking of toilet breaks forces them to ‘hold on’
- Are firms seeing a realised return on AI investment?
- British Firm To Develop Next Gen AI Hardware With Govt. Funding
Comments
Please log in to post a comment.