On the 29th June 2026 - if you searched for Jim Carrey on Google, you would have been presented with a Knowledge Panel stating that he had died the previous day. The panel included a date of death and a biography written in the past tense.
Working in knowledge engineering I saw a failure mode in action on a global scale.
Hadn't heard of AOP before, but funnily enough for a couple of years now I've been calling working with LLMs "aspective coding" - closer to switching context on different features. Will read up a bit more on AOP as it seems like the prior art on this way of thinking.
One example - I'm building an MCP server at the moment for a database I'm working on. In ChatGPT I want to do dry-run posts first that roll back before committing - both are POST requests with a property - and it loves to trigger the safety layer in the tools (for various reasons, it's hard to debug exact causes)
But I think this would make it better - QUERY before POST means different request types, not just the same with a safety flag.
It feels like the market on tokens has changed - until recently you were buying actual usage. Now it feels more like an all-inclusive resource where you buy 'tokens' that the resort decides the value of - watered down drinks and cheap buffet for everyone!
"The factory of the future will have only two employees, a man and a dog. The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment."
I was in New York for a conference 4 years ago, I was discussing with someone a previous project I had worked on in the UK that was a tool for companies to forecast certain risk scenarios "...you know like a building flooding or blowing up"
There were suddenly a lot of unhappy faces looking at me. I guess some folks are still a bit sensitive about that...
I actually thought about this a couple of weeks back that for agents - going backwards actually makes sites more capable - WAP would even be more appropriate. The ultimate irony though is that making websites MORE accessible makes them more agent friendly - the last decade of SPAs is what makes things harder.
Yep, until 1st June 4.6 is still x1 on Copilot, but will jump up quite a bit in coat - 4.7 was already highly priced, and the output was frankly terrible.
It still seems trying to build general models is mostly cost prohibitive - the frontier model provider and resellers are repricing in such a way the return on investment is dropping as developers and users become more cautious of burning their limits.
I'm still of the opinion that models like 4.6 don't need to be improved on - rather they need to be better integrated with more domain specific models in agentic flows.
Reading this, it sounds very like what Ingvar Kamprad did with IKEA - although the structure is slightly different, the intent feels the same.
I work at Inter IKEA, which is actually a franchisor - the structure is a family run foundation sitting at the top, below the group that owns the concept and value chain, then the retailer licence to hold a franchise (https://www.inter.ikea.com/en/this-is-inter-ikea-group/the-i...)
Basically medical science too. My wife was able to diagnose her own anemia that the doctors kept missing, and has since been able to have iron infusions.
The human doctors kept ignoring the signals, kept putting it down to 'diet' and 'exercise' (even though she does plenty of both)
7 years ago we got a Polish Hunting Spaniel, and did our first trip to Poland. Since then we've been back several times, and each time you really see the different - new and upgraded road, city buildings being renovated into new housing and commercial areas - also noticed the costs going up too.
But also you start to notice that definitely a lot of people who left Poland are coming back, and with that skills and new economic opportunities.
I think this is something we'll start to see which is something like a Mandela-Effect, but from LLM results. When we had deterministic search - everyone could see the same result, but now using LLMs knowledge becomes a training and seeding issue. Two people can confidently be given completely different information, so in both cases perceived as true.
Working in knowledge engineering I saw a failure mode in action on a global scale.