DrMikeWatts
@drmikewatts.bsky.social
PhD in artificial intelligence, research experience in IT and ecology. Scientist. Engineer. Educator. Husband. Father.
This is the precedent #AI companies didn't want: www.theguardian.com/technology/2... #ArtificialIntelligence
ChatGPT violated copyright law by ‘learning’ from song lyrics, German court rules
OpenAI ordered to pay undisclosed damages for training its language models on artists’ work without permission
www.theguardian.com
November 17, 2025 at 4:03 AM
This is the precedent #AI companies didn't want: www.theguardian.com/technology/2... #ArtificialIntelligence
#AI give themselves away by being too polite: arstechnica.com/information-... #ArtificialIntelligence
Researchers surprised that with AI, toxicity is harder to fake than intelligence
New “computational Turing test” reportedly catches AI pretending to be human with 80% accuracy.
arstechnica.com
November 17, 2025 at 2:28 AM
#AI give themselves away by being too polite: arstechnica.com/information-... #ArtificialIntelligence
#AI mostly memorise, they don't really learn how to do things: arstechnica.com/ai/2025/11/s... #ArtificialIntelligence
Researchers isolate memorization from problem-solving in AI neural networks
Basic arithmetic ability lives in the memorization pathways, not logic circuits.
arstechnica.com
November 17, 2025 at 12:52 AM
#AI mostly memorise, they don't really learn how to do things: arstechnica.com/ai/2025/11/s... #ArtificialIntelligence
#AI country songs have come a long way since "You Can't Take My Door": www.rnz.co.nz/news/world/5... #ArtificialIntelligence
AI country song rollicks to the top of US sales chart
AI-generated songs are creeping onto the music charts in various scenes.
www.rnz.co.nz
November 16, 2025 at 10:16 PM
#AI country songs have come a long way since "You Can't Take My Door": www.rnz.co.nz/news/world/5... #ArtificialIntelligence
Microsoft says there is no #AI bubble: www.computerworld.com/article/4087... #ArtificialIntelligence
About that 'AI bubble' — Microsoft doesn’t see one.
The general consensus among experts is that the current level of spending on AI technology can’t continue. But for Microsoft, it’s full steam ahead.
www.computerworld.com
November 16, 2025 at 8:26 PM
Microsoft says there is no #AI bubble: www.computerworld.com/article/4087... #ArtificialIntelligence
At the AI Researchers Association conference in Wellington
November 16, 2025 at 7:52 PM
At the AI Researchers Association conference in Wellington
Weekly Review 14 November 2025: computational-intelligence.blogspot.com/2025/11/week... #ArtificialIntelligence
Weekly Review 14 November 2025
Some interesting links that I Tweeted about in the last week (I also post these on Mastodon , Threads , Newsmast , and Bluesky ): When...
computational-intelligence.blogspot.com
November 14, 2025 at 4:13 AM
Weekly Review 14 November 2025: computational-intelligence.blogspot.com/2025/11/week... #ArtificialIntelligence
Enzyme design by #AI-in this case an enzyme to break down and recycle polyurethane: arstechnica.com/science/2025... #ArtificialIntelligence
Neural network finds an enzyme that can break down polyurethane
Given a dozen hours, the enzyme can turn a foam pad into reusable chemicals.
arstechnica.com
November 14, 2025 at 2:26 AM
Enzyme design by #AI-in this case an enzyme to break down and recycle polyurethane: arstechnica.com/science/2025... #ArtificialIntelligence
Energy is now the bottleneck of large scale #AI rollout, not chip shortages: techcrunch.com/2025/11/03/a... #ArtificialIntelligence
Altman and Nadella need more power for AI, but they’re not sure how much | TechCrunch
The CEOs of OpenAI and Microsoft are betting that AI will continue to consume more electricity, but they're not sure how much. That could leave some investors holding the bag.
techcrunch.com
November 14, 2025 at 1:45 AM
Energy is now the bottleneck of large scale #AI rollout, not chip shortages: techcrunch.com/2025/11/03/a... #ArtificialIntelligence
Students fear that #AI is destroying their ability to study, but use it anyway: www.theguardian.com/technology/2... #ArtificialIntelligence
Pupils fear AI is eroding their ability to study, research finds
One in four students say AI ‘makes it too easy’ for them to find answers
www.theguardian.com
November 14, 2025 at 12:39 AM
Students fear that #AI is destroying their ability to study, but use it anyway: www.theguardian.com/technology/2... #ArtificialIntelligence
Developer roles for #AI agents are less in demand as AI takes over more roles: www.theregister.com/2025/11/03/d... #ArtificialIntelligence
Agentic AI dev roles are less in demand as agents take over
: IEEE survey of senior techies in six countries finds recrutiment for data analytics, and machine learning on the up
www.theregister.com
November 13, 2025 at 10:32 PM
Developer roles for #AI agents are less in demand as AI takes over more roles: www.theregister.com/2025/11/03/d... #ArtificialIntelligence
Massive amounts of money are being poured into #AI, mostly for data centres to run them: www.theguardian.com/global/2025/... #ArtificialIntelligence
The mind-boggling valuations of AI companies
You try wrapping your head around a string of deals worth nearly $600bn
www.theguardian.com
November 13, 2025 at 8:40 PM
Massive amounts of money are being poured into #AI, mostly for data centres to run them: www.theguardian.com/global/2025/... #ArtificialIntelligence
AI clones of the dead are improving but still tasteless: www.rnz.co.nz/news/on-the-...
Can you really talk to the dead using AI? We tried out 'deathbots' so you don't have to
'Deathbots' are AI systems designed to simulate voices, speech patterns and personalities of the deceased.
www.rnz.co.nz
November 13, 2025 at 4:33 AM
AI clones of the dead are improving but still tasteless: www.rnz.co.nz/news/on-the-...
LLM AI still can't explain their internal reasoning: arstechnica.com/ai/2025/11/l...
LLMs show a “highly unreliable” capacity to describe their own internal processes
Anthropic finds some LLM “self-awareness,” but “failures of introspection remain the norm.”…
arstechnica.com
November 13, 2025 at 3:42 AM
LLM AI still can't explain their internal reasoning: arstechnica.com/ai/2025/11/l...
Large scale #AI requires large scale nuclear: www.theregister.com/2025/10/31/d... #ArtificialIntelligence
Datacenter biz and nuke firm join forces for Texas AI ranch
: The bit barn will run on gas power first.
www.theregister.com
November 13, 2025 at 2:30 AM
Large scale #AI requires large scale nuclear: www.theregister.com/2025/10/31/d... #ArtificialIntelligence
The latest copyright ruling on #AI does not really set a useful precedent: www.theregister.com/2025/11/04/u... #ArtificialIntelligence
Getty loses UK copyright battle against Stability AI
: Experts disagree about what the ruling means for AI training on copyrighted material
www.theregister.com
November 13, 2025 at 12:29 AM
The latest copyright ruling on #AI does not really set a useful precedent: www.theregister.com/2025/11/04/u... #ArtificialIntelligence
UK court rules #AI does not store copyrighted images: www.theguardian.com/media/2025/n... #ArtificialIntelligence
AI firm wins high court ruling after photo agency’s copyright claim
Ruling in case brought by Getty Images against Stability AI is seen as a blow to copyright owners
www.theguardian.com
November 12, 2025 at 11:51 PM
UK court rules #AI does not store copyrighted images: www.theguardian.com/media/2025/n... #ArtificialIntelligence
Yet another #AI model is yanked from public use after it hallucinates facts, this time about a US senator: www.theverge.com/news/812376/... #ArtificialIntelligence
Google pulls AI model after senator says it fabricated assault allegation
Senator Marsha Blackburn accused Google of anti-conservative bias.
www.theverge.com
November 12, 2025 at 10:32 PM
Yet another #AI model is yanked from public use after it hallucinates facts, this time about a US senator: www.theverge.com/news/812376/... #ArtificialIntelligence
It looks like most tests for #AI safety are fundamentally flawed: www.theguardian.com/technology/2... #ArtificialIntelligence
Experts find flaws in hundreds of tests that check AI safety and effectiveness
Scientists say almost all have weaknesses in at least one area that can ‘undermine validity of resulting claims’
www.theguardian.com
November 12, 2025 at 9:34 PM
It looks like most tests for #AI safety are fundamentally flawed: www.theguardian.com/technology/2... #ArtificialIntelligence
A decentralised #AI, that doesn't need a data centre: www.theregister.com/2025/11/02/f... #ArtificialIntelligence
Fortytwo has the answer to everything: decentralized AI
: No datacenters required
www.theregister.com
November 12, 2025 at 8:49 PM
A decentralised #AI, that doesn't need a data centre: www.theregister.com/2025/11/02/f... #ArtificialIntelligence
Microsoft apologises for its confusing pricing around its #AI offerings: www.stuff.co.nz/nz-news/3608... #ArtificialIntelligence
Microsoft apologises, offer refunds after AI price hike
Microsoft in New Zealand and Australia have apologised and offered refunds to customers for the way it charged customers to use its AI tools.
www.stuff.co.nz
November 12, 2025 at 3:38 AM
Microsoft apologises for its confusing pricing around its #AI offerings: www.stuff.co.nz/nz-news/3608... #ArtificialIntelligence
Which #AI markets are dominated by current players, and which are still open: techcrunch.com/2025/11/03/e... #ArtificialIntelligence
Elad Gil on which AI markets have winners — and which are still wide open | TechCrunch
Over the last year, certain AI markets appear to be nearly sewn up by startup market leaders. But beyond these areas, a vast swath of AI remains anyone’s game.
techcrunch.com
November 12, 2025 at 2:23 AM
Which #AI markets are dominated by current players, and which are still open: techcrunch.com/2025/11/03/e... #ArtificialIntelligence
#AI browsers can do some useful things, but cannot be trusted: www.computerworld.com/article/4083... #ArtificialIntelligence
AI web browsers are cool, helpful, and utterly untrustworthy
Maybe it's time we rethink just how much we're depending on AI these days, before it blows up in our faces. Just saying!
www.computerworld.com
November 11, 2025 at 10:31 PM
#AI browsers can do some useful things, but cannot be trusted: www.computerworld.com/article/4083... #ArtificialIntelligence