Plus: JPMorgan leads Anthropic's retail IPO push; Google DeepMind ships Genie 4 with real-time audio and steering
Welcome in, today鈥檚 issue asks us to look a little closer at the claims we鈥檝e been taking for granted.
RESEARCH
馃攳 Audit finds serious flaws in AI education studies
Image source: hechingerreport.org
There are 39 meta-analyses reporting substantial benefits for AI in education. A forensic audit of 14 of the 39 meta-analyses found serious conceptual and statistical flaws.
The details:
One meta-analysis was retracted almost a year after its May 2025 publication in a Nature journal.
Patrick O'Neill found recurring patterns of errors in the meta-analyses.
Susanna Loeb's team collected almost 900 studies on AI in education and determined that only 20 were of high enough quality for a recent summary.
Anthropic has selected JPMorgan Chase to lead U.S. retail share sales for its planned IPO. Anthropic is considering allocating 15% to 20% of its IPO shares to retail investors.
The details:
JPMorgan CEO Jamie Dimon is expected to participate in the bank's retail investor outreach campaign.
Citigroup is set to coordinate the international retail offering for Anthropic's IPO.
Anthropic filed publicly for an IPO on Monday, with Goldman Sachs and Morgan Stanley leading the offering.
馃實 Google DeepMind ships Genie 4 with real-time audio and steering
Image source: techcrunch.com
Google DeepMind released Genie 4, its newest world model, in research preview on Tuesday. Genie 4 generates real-time audio that matches the simulated world, including music, ambient noises, and interactive dialogue.
The details:
Genie 4 can maintain a consistent scene for up to two minutes, double its predecessor鈥檚 limit.
Genie 4 allows users to type prompts mid-simulation to change the scene, such as altering weather or camera angle.
In a research experiment, a Gemini 4 Argon model broke down tasks into steps that Genie 4 simulated in real time.
