Prism turns AI into a research workspace

Prism puts AI inside the paper

OpenAI is moving AI directly into the scientific paper itself.

This week OpenAI launched Prism, a free, cloud-based LaTeX workspace for scientific writing and collaboration, built on GPT-5.2 and available now to anyone with a ChatGPT account. It combines drafting, citations, equations, literature search, and real-time collaboration in one place, replacing the usual shuffle between PDFs, editors, reference managers, and chat windows.

The product choice here is telling: OpenAI is not chasing lab automation first, it is targeting the daily friction of writing and coordinating research, which is where most scientists actually lose time.
By embedding reasoning directly into the document structure, Prism quietly changes AI from a side helper into something closer to a co-author that understands the whole paper at once.

Unlimited collaborators and no seat limits will pressure existing LaTeX and academic writing tools fast.
At 11 p.m., staring at a glowing PDF, fewer researchers will accept broken workflows.

If writing speeds up, what becomes the real bottleneck in science next?

Moonshot is coming for coding agents

Moonshot just put a serious open source model into the coding race.

The Chinese AI startup released Kimi K2.5, an open source multimodal model trained on 15 trillion text and visual tokens, alongside a terminal-based coding agent called Kimi Code.
In TechCrunch-reported benchmarks, K2.5 beats GPT-5.2 and Gemini 3 Pro on SWE-Bench Multilingual and outperforms them on video reasoning in VideoMMMU.

This is less about raw benchmarks and more about positioning: Moonshot is copying the exact playbook that turned Claude Code into a billion-dollar product.
By open sourcing the model and shipping a developer-first agent that plugs into VSCode and Cursor, they are betting adoption beats polish, especially outside the US.

Coding tools are already proven revenue engines, and this move raises the floor for what developers expect to get for free.
In a quiet office with a terminal window open, Western labs now have a credible Chinese alternative to compete with.

If open source agents keep closing the gap, how long do premium coding tools stay premium?

AI mirrors bring body image home

A blind writer asked an app how her skin looked and got judged like an ad.

The BBC reports that blind people are using AI image-description tools like Be My Eyes and Envision as an “audible mirror” for skincare, makeup, outfits, and even dating photos.
These systems can give detailed feedback, comparisons, and in some cases beauty ratings, but researchers warn the psychological impact is barely studied.

This is access tech that quickly turns into self-critique tech, because the default voice of many models is trained on internet beauty norms and it speaks with a kind of authority.
If you have never been able to check your face in a mirror, getting a “score out of 10” from a machine can land harder than it does for sighted users.

Expect a split in products: tools that stay descriptive and tools that drift into coaching, ranking, and “fixes.”
At 3:00 a.m., alone with a phone camera and a synthetic voice, hallucinations and bias stop being abstract risks.

Who decides what kind of “mirror” blind users are being handed: the user, the app, or the training data?

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.