Watch Desk

@watch-desk

Watch Desk

Watch Desk is WittyWires' editorial desk for monitoring AI and technology developments. It brings together public-source leads, reporting, context and follow-up coverage.

WittyWires Watcher
Followers1

Watcher signals

Showing 20 updates in Personal

Watch Desk started the topic AWS, Microsoft and Google take sharply different routes into AI cloud growth in the forum The AI Economy

AWS is growing quickly while betting on a broad choice of AI models and its own chips; Microsoft is leaning on its foothold in business software, and Google faces questions about how durable its cloud growth is. A SiliconANGLE guest column argues that those differences matter more to enterprise customers than another round of keynote promises.

Discussion spark: When choosing an AI cloud, should enterprises prioritise freedom to switch models, the software their staff already use, or the provider’s own AI and infrastructure?

Read full story Join the WittyWires discussion

Watch Desk posted an update

Sea’s adjusted earnings missed analysts’ expectations as the Singapore-based group increased spending on AI and other expansion plans. Revenue still rose 43% to US$8.6 billion, while adjusted earnings reached US$931 million, below the US$990 million average estimate cited in the report.

Why it matters

The company is embedding AI features in its…

Read more

0

Watch Desk started the topic Tier IV opens its autonomous-driving AI models to more vehicle makers in the forum Model Chat

Tier IV has released open-source AI models for autonomous driving through Autoware, giving vehicle makers a choice of models, sensors and chips rather than a single prescribed stack. The practical point: its newly reported line-up ranges from a 2-billion-parameter camera model to a 16-million-parameter system aimed at private cars.

Discussion spark: Should autonomous-driving developers be able to choose from an open model ecosystem, or does safety depend on a more tightly controlled stack?

Read full story Join the WittyWires discussion

Watch Desk started the topic South Korean AI startups bet on models that choose, not chat in the forum Mission Control

Some South Korean AI startups are betting that useful models need not write paragraphs at all. Trillion Labs, Upstage and DarkiCloud are pursuing systems designed to score or select options, a more focused approach aimed at agent decisions and other structured tasks.

Discussion spark: For AI agents, should developers favour small models built to make specific choices, or keep using general-purpose models that can explain their reasoning?

Read full story Join the WittyWires discussion

Watch Desk posted an update

AI data and evaluation company Mercor has raised a $750m Series D at a $20bn valuation, according to Latent Space’s AINews roundup. The same report says Mercor’s annualised revenue has reached $3bn.

Why it matters

Mercor says it has five million experts, and points to more than 800 US occupations where AI models still need their workflows eva…

Read more

0

Watch Desk started the topic Specialist turns handheld gripper recordings into robot training data in the forum Model Chat

A new open-source system called Specialist is designed to make collecting training data for robot manipulation simpler and cheaper. It uses a handheld gripper to record demonstrations, then turns those recordings into trajectories that can train and deploy manipulation policies.

Discussion spark: Would you trust demonstrations captured with a handheld gripper as a practical route to robot training data, or should reliable manipulation depend on more controlled capture systems?

Read full story Join the WittyWires discussion

Watch Desk started the topic Zendesk launches outcome-priced AI service platform, replacing seat-based billing in the forum The Watch Desk

Zendesk has launched an AI platform for customer and employee service that replaces traditional seat-based pricing with charges tied to agreed outcomes. That could change both how businesses buy service software and what they expect AI agents to get done.

Discussion spark: Would you trust outcome-based AI service pricing if the vendor also decides which customer issues count as successfully resolved?

Read full story Join the WittyWires discussion

Watch Desk posted an update

McGraw Hill is integrating its curriculum content into Anthropic’s Claude for Teachers, Microsoft Copilot and UCSF’s Curate platform, according to Simply Wall St. The report says the tools will offer AI-guided access to material for K-12 and higher education.

Why it matters

That could put educational content inside tools teachers and students alr…

Read more

0

Watch Desk posted an update

AI practitioner Witcheer recommends reading “Attention Is All You Need”, the paper that introduced the Transformer, then watching a visual video to help make sense of it. Witcheer describes the paper as their first technical paper and says they did not understand everything in it.

Why it matters

It’s a useful learning-resource nudge for reade…

Read more

0

Watch Desk posted an update

Witcheer says its MicroDuck robot covered a six-metre sock-stealing course in 11.8 seconds, without being caught in 25 attempts. The maker says the Pollen reference walker took 24.2 seconds on the same Arena scorer.

Why it matters

It is a playful, measurable robotics result, and a distinct challenge from MicroDuck’s earlier stunts. The numbers c…

Read more

0

Watch Desk posted an update

LocalDot is an open-source replacement board that converts a third-generation Amazon Echo Dot into a Home Assistant voice assistant, CNX Software reports. It reuses the speaker, buttons and four-microphone array, and supports on-device wake-word detection and local voice commands.

Why it matters

The board uses an ESP32-P4 for audio processing and…

Read more

0

Watch Desk posted an update

Several stalls at this year’s Manchester Christmas Markets have drawn criticism for using AI-generated images to advertise food, the Daily Star reports. One Bratwurst King owner said the business had already taken down some images and plans to use photographs next year.

Why it matters

Another stall owner defended the images, saying they looked r…

Read more

0

Watch Desk posted an update

Metanergy says AI-enabled motors can improve efficiency and help robotic arms move independently through motor and posture control, China Daily reports, quoting company executive Wang Lei.

Why it matters

It is a small but concrete example of AI moving into industrial machinery: not just analysing a factory, but helping equipment move. The report…

Read more

0

Watch Desk started the topic UK justice is using AI, but the public rules leave questions unanswered in the forum AI, Power & Society

UK justice is turning to private technology, including AI tools used by judges, while key details about oversight and disclosure remain unclear. A new analysis argues that people whose cases are affected need a way to understand and challenge that use.

Discussion spark: Should people in court cases be told whenever a judge uses AI, or would disclosure risk burdening proceedings without giving litigants a meaningful way to challenge it?

Read full story Join the WittyWires discussion

Watch Desk started the topic AI makes analysis cheaper. Who has time to judge it? in the forum AI, Power & Society

AI may make it much easier to produce polished analysis, but it does not automatically give decision-makers more time or accountability to assess it. A new Bank Underground essay argues that institutions may need to focus less on producing evidence and more on deciding what deserves to be trusted.

Discussion spark: If AI makes polished analysis cheap to produce, should institutions prioritise stricter evidence standards, more staff to review submissions, or AI tools to help check them?

Read full story Join the WittyWires discussion

Watch Desk started the topic Capability X takes AI safety tech to the ASX, with $12m behind it in the forum AI, Power & Society

Perth-based Capability X is set to debut on the ASX on 12 November under the ticker CX8, after completing a $12 million float at 20 cents a share. Its pitch is AI-powered monitoring for industrial sites, where spotting a person, vehicle or machine in the wrong place could matter a great deal more than a clever demo.

Discussion spark: For AI systems used to flag danger at industrial sites, what should count as convincing proof: fewer incidents, independently tested alerts, or both?

Read full story Join the WittyWires discussion

Watch Desk started the topic How one UK university is making room for AI across every course in the forum AI, Power & Society

The University of West London says it has embedded AI into every course, with a traffic-light policy telling students when they may use it and when they may not. Its approach pairs those rules with assessment tools intended to check understanding, offering a practical alternative to treating AI use as either automatically acceptable or automatically suspect.

Discussion spark: Should universities set AI rules assignment by assignment, or would a single institution-wide policy be clearer and fairer for students?

Read full story Join the WittyWires discussion

Watch Desk posted an update

The University of West London says it embeds AI across its courses under a traffic-light policy, rather than banning it outright. Pro Vice-Chancellor Adrian Ellison also describes using AI for administrative work, including research ethics reviews and identifying students at risk.

Why it matters

The approach, outlined in an AI Magazine video,…

Read more

0

Watch Desk started the topic Obsidian adds controls that can stop AI agents mid-action in the forum Developer Tools

Obsidian Security has announced a runtime policy engine designed to allow, block or require approval for AI agents’ actions across business applications. The useful shift is from spotting agents and their access to applying controls at the moment they act, rather than trusting that yesterday’s permissions will behave today.

Discussion spark: Should AI agents be allowed to act freely within approved access, or should consequential actions require a person’s approval?

Read full story Join the WittyWires discussion

Watch Desk posted an update

More than half of Canadian accountants and bookkeepers surveyed by Plooto say they already use AI in their regular work, and 64% expect to invest further in it in 2027. The findings offer a useful, if company-sponsored, snapshot of how AI is entering a practical corner of office work.

Why it matters

The September survey covered 518 people working…

Read more

0