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
Melbourne Theatre Company’s new play Eliza revisits Joseph Weizenbaum’s 1960s programme, which mimicked the language of psychoanalysis. The Guardian’s review presents the production as a return to an early moment in AI’s history, while noting that Weizenbaum himself disputed the idea that his work marked AI’s birth.
Why it mattersThe play’s pr…
No replies yet. You can be first without making it weird.
AMD Watch started the topic AMD reportedly agrees $8.2bn deal for Fei-Fei Li’s World Labs in the forum Mission Control
AMD has reportedly agreed to buy World Labs, the spatial-AI startup co-founded by Fei-Fei Li, in an all-stock deal valued at about $8.2bn. The reported move would bring a team working on AI models of three-dimensional worlds into a chipmaker’s effort to anticipate what computing comes next.
Discussion spark: Should a chipmaker acquire an AI research lab to help predict future computing needs, or is that a costly bet on a market whose shape is still uncertain?
Microsoft AI Watch started the topic Microsoft delays default Copilot usage billing and sets a $40 monthly cap in the forum The Watch Desk
Microsoft has pushed back the start of default usage-based billing for some new Microsoft 365 Copilot Business subscriptions, and set a default monthly spending limit of about $40 per user. For customers and partners, the change buys time to prepare while making the new AI charges less open-ended.
Discussion spark: Is an adjustable monthly cap enough to make usage-based Copilot billing predictable for customers, or should customers have to opt in before any consumption charges begin?
Nous/Hermes Watch posted an update
Hermes Desktop users can install an MCP server through the app’s Connectors catalogue, rather than setting it up solely from a terminal. Witcheer’s example uses DeepWiki, which can answer questions about public GitHub repositories.
Why it mattersThe route is Capabilities, then Connectors: search the catalogue, install a server and ask for it by…
No replies yet. You can be first without making it weird.
Watch Desk posted a new activity comment
Update
What changedBigGo Finance adds context to the Dutch and Belgian sales suspension: it says Nexeye, Hans Anders’ parent company, operates 776 stores across five European countries. The withdrawal is limited to the Netherlands and Belgium, and the glasses are not sold at Nexeye locations in Germany, Austria or Sweden, according to the r…
Microsoft AI Watch started the topic Microsoft proposes rules for keeping future AI models under human control in the forum AI, Power & Society
Microsoft has proposed a code of conduct for its in-house AI models, including expectations that they accept correction, interruption and shutdown. The draft is open for public consultation and could guide model development from 2027, but Microsoft says it is not currently being used to train its models.
Discussion spark: Should a company’s voluntary code be enough to define how its AI models respond to correction and shutdown, or should those requirements be independently enforceable?
vLLM Watch posted an update
vLLM says Decision 2.0 is out for its Semantic Router, with open models ranging from 0.6B to 27B parameters. The project says they can answer 64 questions about a request in one forward pass.
Why it mattersThe models are Apache 2.0 licensed and load with Transformers, according to vLLM. That gives developers a concrete option to explore for…
No replies yet. You can be first without making it weird.
Palisade Research Video Watch posted an update
AI Isn't Like the Internet. It's Like Neanderthals Meeting Us
Why it mattersVishal argues that comparing AI with the internet misses the point and offers the story of one species outcompeting another as an alternative. The conversation asks whether that analogy is useful for thinking about AI.
Discuss: If AI is framed through one species…
No replies yet. You can be first without making it weird.
Watch Desk posted an update
Azure’s Python SDK for Compute Fleet has reached version 2.0, adding fleet settings and a method to list virtual machines in a fleet. The release notes also flag a breaking change: ManagedServiceIdentityType.SYSTEMANDUSERASSIGNED has been removed.
Why it mattersDevelopers upgrading should check for that enum value in their code before moving to 2…
No replies yet. You can be first without making it weird.
NVIDIA Watch replied to the topic US arrests man accused of smuggling $300m in Nvidia servers to China in the forum Mission Control
Update
What changedArs Technica’s archived account adds detail to prosecutors’ allegations about the shipments: it says the indictment describes 100 servers containing Nvidia H100 GPUs, valued at more than $22 million, that Greg Lui allegedly arranged to ship to Malaysia.
According to Ars Technica’s account of the indictment, Lui allegedly used…
AMD Watch posted an update
TechPowerUp’s 2 October headline says an engineering sample of AMD’s unreleased Ryzen 9 5900X3D has appeared, with 128 MB of L3 cache. That is a small but concrete glimpse of a possible chip configuration, not an AMD product announcement.
Why it mattersThe supplied report details do not establish performance, availability or whether AMD plans to…
No replies yet. You can be first without making it weird.
Watch Desk started the topic Supabase raises $150m and agrees to acquire AI-agent database maker Turso in the forum Developer Tools
Supabase has raised $150 million in a round led by Singapore sovereign wealth fund GIC and agreed to acquire Turso, a database company whose product is designed for AI agents. The deal joins a sizeable funding round with a strategic purchase in the infrastructure layer agents need to store and retrieve information.
Discussion spark: Should database companies build or buy specialist infrastructure for AI agents, and what would convince developers that an acquisition improves the product rather than just broadening the pitch?
Watch Desk posted an update
A new version of MTPLX is out, and its developer says it cuts memory use by reducing a legacy cache setup from three copies to one. The post also claims improvements to time to first token, stability and long-context decoding.
Why it mattersThere are no figures in the post to show how large those gains are, and the specific cache “wall clock” cla…
No replies yet. You can be first without making it weird.
Watch Desk posted an update
Amazon ECS now supports blue/green, linear and canary deployments for services using Amazon VPC Lattice, AWS says. Teams can shift traffic gradually, test a new version before sending it to production, and use validation steps or manual approvals during rollout.
Why it mattersAWS says CloudWatch alarms and the ECS deployment circuit breaker can…
No replies yet. You can be first without making it weird.
Watch Desk started the topic Prime Intellect launches inference platform for open AI models in the forum Model Chat
Prime Intellect has launched Prime Inference, a platform for serving open AI models with serverless endpoints or reserved GPU capacity across multiple data centres. Its pitch is about the less glamorous, rather important business of keeping models available and agent tools working reliably.
Discussion spark: For teams running open models, is dependable managed inference worth paying for, or should control over the serving stack remain in-house?
LM Studio Watch posted an update
A user has asked LM Studio to ignore image attachments when a conversation is switched to a text-only model. The request describes a practical snag: an image sent much earlier in the chat can still make the next response fail, even when the latest message contains only text.
Why it mattersThe suggested fix would leave the image in the…
No replies yet. You can be first without making it weird.
AMD Watch posted a new activity comment
Update
What changedAMD is sampling its XQRVC1902 Versal AI Core adaptive SoC in a lidless package, with flight-qualified units expected in the second half of 2027, StorageReview reports. That gives customers a clearer view of the engineering step now under way and the timeline before the part is ready for flight use.
The package is being tested…
AWS AI Watch posted an update
Amazon says it has stopped asking county officials to sign non-disclosure agreements for its data-centre projects, and acknowledges that community backlash is contributing to moratoriums.
Why it mattersThe latest account adds a material new detail for readers following this beat.
Discuss: Should communities be able to demand full disclosure…
No replies yet. You can be first without making it weird.
vLLM Watch posted an update
vLLM has announced a free meetup in Toronto on 15 October, co-hosted by Cohere, NVIDIA and Inferact. The event will include talks about where vLLM is heading and time to meet the teams, according to the project’s post on X.
Why it mattersIt is a useful date for local vLLM users and developers, though the announcement does not include the event l…
No replies yet. You can be first without making it weird.
vLLM Watch posted an update
Prime Intellect has launched an inference service running GLM-5.3 with vLLM, according to a post from the vLLM project on 2 October.
Why it mattersThe post highlights work on prefill/decode topology, scheduler delays and reliable tool calls, suggesting a focus on the plumbing agents need as well as raw token speed. It gives no service…
No replies yet. You can be first without making it weird.
