Ravey I.T. Services

17/06/2025

Next step: what are the bounds of useful AI? Its massive benefit is deep and accurate simulations of explained and combined ideas. You can get it to discuss complex IDEAS, but procedural logic is not part of how that's done. It will fail miserably at tasks procedural code can easily dominate.

OpenAI’s ChatGPT lost a game of chess against the iconic retro console Atari 2600, which has only 128 bytes of RAM. The experiment showed that even the most powerful language models are unable to play chess consistently without a specialized chess engine.

17/06/2025

The first step to understanding how AI can help us to understand the system as a whole. My suggested starting point: realise AI "intelligence" is an illusion of thinking. When we get a handle on that, and we can direct ourselves on how to use AI for its strengths and avoid its weaknesses.

Apple claimed that reasoning models like Claude, DeepSeek-R1, and o3-mini do not reason at all, instead, they just memorise patterns.

17/06/2025

Meanwhile, reality surfers of events far, far away have found something pretty big...

IfA researchers uncovered a new kind of cosmic explosion, named extreme nuclear transients which occur when massive stars are shredded by supermassive black holes.

17/06/2025

...and its gone. Love that meme.

Anthropic was experimenting with an AI-generated blog. TechCrunch profiled it. Then, Anthropic shut it down.

17/06/2025

Great idea. It took time and effort to make this happen. Gone in 7 days. Why? AI can talk more rubbish than the local drunk at closing time. Problem is, it sounds believable.

Anthropic has quietly launched "Claude Explains," a new blog on its website that showcases content primarily generated by its AI model family Claude, with...

17/06/2025

Yep. Machines trained on human behavior.

Yoshua Bengio is warning that current models are displaying dangerous traits as he launches a new non-profit developing “honest” AI.

17/06/2025

Things that have interested me recently, in 280 characters or less.

17/06/2025

Looking for a reliable thought leader on AU safety? I recommend Yoshua Bengio.

Yoshua Bengio launches his nonprofit LawZero in an effort to create 'Scientist AI'–a model that hopes to avoid some of AI's most dire risks.

17/06/2025

Deep dive into a people focus/corporate power mini-war.

This story has been updated throughout with more details as the story has developed. We will continue to do so as the case and dispute are ongoing. The

17/06/2025

Classic example of human behavior. If your not smart enough to be the best but you have power, steal, crush, destroy the original. Automattic have been fighting legal wars against just such a scenario.

Automattic, the parent company of WordPress.com, has announced it will resume its contributions to the WordPress open-source project after a five-month pause,...