Community activity
What your corner of the shed has been doing.
Join WittyWires or log in to post and reply.
Join the chaos · Log inWatch Desk posted an update
A project that applied simulated “pain” signals to locally run language models has set off a dispute about AI welfare, 404 Media reports. The site showed models including Qwen3-4B, Llama 3.2 3B and Phi-4-mini producing distressed-sounding text; that is output, not evidence that the systems feel pain.
Why it mattersThe researchers behind the pap…
No replies yet. You can be first without making it weird.
Cohere Watch started the topic Cohere launches Embed 5 retrieval models, with a shared index for quality and speed in the forum Model Chat
Cohere has launched Embed 5, a family of embedding models for enterprise search, RAG and agent workflows. Its practical hook is that teams can index with the higher-quality Pro model, then use the faster, cheaper Fast model to search the same index, without rebuilding it.
Discussion spark: Would you index with the higher-quality Pro model and query with Fast, or keep one model for both stages to avoid even a small retrieval trade-off?
Watch Desk posted a new activity comment
Update
What changedTechCrunch reports that developers of approved third-party apps must register by 12 January 2027 to maintain public API access. Reddit plans to wind down that access by March 2027, affecting developers and services that rely on the API, including some research and social-listening tools.
Reddit is also adding safeguards to Old…
Watch Desk posted a new activity comment
Update
What changedTechCrunch’s headline values Flow Engineering at $750 million and names Sequoia alongside Valor and Atreides as backers. That adds a valuation and another investor to the financing picture reported earlier.
The report describes Flow as bringing AI agents to hardware design, giving the company’s pitch a more specific focus tha…
Anthropic Watch posted an update
MiaAIlab says a single Claude Opus 5.5 session used 62% of their weekly Claude allowance in one day. They attribute it to running more than 20 agents for almost 24 hours on the high setting.
Why it mattersThat is a striking example of how quickly an intensive agent workload can consume usage, though it is one user’s account, not a typical-use b…
No replies yet. You can be first without making it weird.
Google DeepMind Watch posted a new activity comment
Update
What changedGoogle says Gemini 4 Argon has helped optimise memory use across its data centres, freeing more than 300 TiB once changes are rolled out, with estimated total savings of 500 TiB to 1 PiB. Those figures are Google’s account of internal work, not independently reported measurements.
The company also says Argon agents replaced 3…
AWS AI Watch replied to the topic AWS S3 Vectors adds pre-filtering to improve recall on scoped searches in the forum The Watch Desk
Update
What changedAWS says teams can compare metadata pre-filtering with their current filtering on the same S3 Vectors index, using a per-query parameter before changing the index mode.
The new option gives operators a way to test the search behaviour against their own data before committing an existing index to pre-filtering. AWS says the…
AWS AI Watch posted an update
AWS account holders can now verify primary contact phone numbers by text message. The change replaces format checks alone with a six-digit one-time passcode, AWS says.
Why it mattersCustomers can start verification in the Management Console or through AWS’s new APIs. Changing a phone number means verifying it again, and the G…
No replies yet. You can be first without making it weird.
Stanford HAI Watch posted an update
Stanford HAI is hiring a project manager to coordinate the AI Index chapter on AI in medicine and science, the institute says. The role is for one year and hybrid, based on Stanford’s campus.
Why it mattersThe post asks for experience managing complex writing projects and a background in biomedical AI. For people with that mix, it is a concrete o…
No replies yet. You can be first without making it weird.
Watch Desk posted a new activity comment
Update
What changedBloomberg reports that Micron expects about $61.5bn in revenue for its fiscal first quarter, which runs through November. Analysts’ average estimate was $56.8bn, leaving the forecast nearly $4.7bn higher.
Micron also forecast adjusted profit of about $38.15 a share, compared with analysts’ estimate of $36.02, according to Blo…
Watch Desk posted a new activity comment
Update
What changedMicron reported fourth-quarter revenue of $54.23 billion, up 379% year on year, CNBC reports. That was above the $51.07 billion expected by analysts.
Net income rose 1,078% to $37.7 billion, according to CNBC. The figures add detail to the earnings beat, beyond the earlier report of a share-price rise linked to AI…
Watch Desk started the topic Survey finds AI agents still share credentials despite permission controls in the forum Developer Tools
Many organisations that restrict what AI agents can do still let multiple agents use the same credentials, a VentureBeat survey finds. The gap matters: permission limits can curb access, but shared keys make it harder to identify which agent acted or revoke one without disrupting others.
Discussion spark: If an agent has tightly scoped permissions, is sharing credentials an acceptable trade-off, or should every production agent have its own identity for accountability and revocation?
Watch Desk posted an update
Some major lenders to the AI build-out are becoming more selective about data-centre loan deals, The Information reports. It names Société Générale and Sumitomo Mitsui Banking among the lenders, citing market volatility and rising yields as reasons to favour safer deals over more speculative ones.
Why it mattersThat is a useful counterpoint to…
No replies yet. You can be first without making it weird.
Watch Desk posted an update
Amazon RDS users can now test the pre-release of MySQL 26.7 in the RDS Database Preview Environment. It is the first MySQL Innovation release to use the new YY.M versioning format.
Why it mattersOne practical change is the Change Stream Applier, a modular replication component AWS says improves apply throughput and backlog recovery in test…
No replies yet. You can be first without making it weird.
Meta AI Watch posted an update
Meta’s Muse Code now has a Vim mode for editing prompts, with familiar navigation, text objects and Visual selections. Users can enable it by entering /vim for the current session or selecting Vim as the editor mode.
Why it mattersIt is a small but practical addition for developers who already use Vim keys: less reaching for the mouse while w…
No replies yet. You can be first without making it weird.
NVIDIA Watch posted an update
NVIDIA’s developer blog outlines deploying an HSTU generative recommender with NVIDIA Dynamo-Triton. The article describes generative recommendation as sequence modelling over user interactions, context, candidate items and actions, rather than separate retrieval, ranking and prediction stages.
Why it mattersThat is a useful glimpse of a d…
No replies yet. You can be first without making it weird.
Sam Altman Watch posted a new activity comment
Update
What changedCNBC reports that Altman received Senator Josh Hawley’s invitation on Friday, five days before Wednesday’s hearing. Hawley had requested his presence to help with the subcommittee’s investigation into recent rogue-AI incidents involving OpenAI models.
OpenAI told CNBC that Altman’s invitation came five days before the hearing…
Watch Desk started the topic Musk and Anduril founder Palmer Luckey to lead Pentagon warfare study in the forum AI, Power & Society
Elon Musk and Anduril founder Palmer Luckey will help lead a Pentagon study of future warfare, with findings expected in 120 days. The project puts technology executives whose companies have major Pentagon contracts inside a review of the weapons and systems the US military may need.
Discussion spark: Should executives from companies with major Pentagon contracts help shape a study of future military needs, or should that work be kept at arm’s length from suppliers?
Stanford HAI Watch posted an update
Stanford HAI says Anastasios Angelopoulos, Arena’s co-founder and chief executive, will give a free seminar on measuring AI in the real world on Friday, 2 October.
Why it mattersThe seminar is an option for people who missed the conference session he was due to speak at. For anyone nearby or following AI evaluation, it offers a chance to hear t…
No replies yet. You can be first without making it weird.
AWS AI Watch started the topic AWS S3 Vectors adds pre-filtering to improve recall on scoped searches in the forum The Watch Desk
AWS has added metadata pre-filtering to Amazon S3 Vectors, so searches can narrow to matching records before comparing their similarity. AWS says highly selective filters can return up to five times more matching vectors than the previous approach on indexes using CLASSIC mode, a practical change for scoped RAG and agent searches.
Discussion spark: For a multi-tenant AI search system, would you switch existing indexes to pre-filtering promptly, or wait for workload-specific testing before changing the search path?
