I’m not sure about that myself. But it would definitely be nice if we had more reliable ways of measuring model performance. Also feels a good time to plug this again: open.substack.com/pub/asimovad...
Reddit comment describing a theory that Anthropic’s models are getting worse because of their economic incentive to increase efficiency.
Evidence is mostly anecdotal, or that it’s failing the ‘should I walk to the car wash’ style prompts. So this might be a kind of collective delusion, but there are some pretty interesting takes floating around about why nerfing models might be in Anthropic’s interests.
People are convinced 4.7 is a downgrade on 4.6. It seems “lazier” or like it’s “not thinking” anymore (quotes from Reddit).
Watching the fallout from Anthropic’s latest model (Opus 4.7) release has been fascinating
@simonpoulton.com thinks that there is a "possibility that shopping agents empower consumers so they aren’t forced onto a platform to transact.” I think this exemplifies exactly why LLMs might only get worse from here asimovaddendum.substack.com/p/are-llms-t...
The technology just feels momentous. But, as I argue here, no one has worked out how to make it profitable: substack.com/@asimovsadde...
This is shocking considering that ChatGPT got 100 million users in 2 months.
Thanks Jack! Trying my best to put all that I learnt in HPSC0061 to good use :)
Love this. Every time I use Google Scholar it amazes me to think that it is how Google once was. I miss the point of search being... search.
Yet the lack of profits at these Al companies has an upside for the user - at least for now. Searching for things like candles on ChatGPT is so unusually pleasant at the moment - particularly compared to Amazon and Google - precisely because its search function has not yet been warped by the pressure to generate the kind of profits needed to recoup past losses. However, if these Al tools are to survive financially, the economic reality is that they will have to change.
Provocative title here from @rufusrock.bsky.social but it's hard to disagree on this point 1/ pasimovaddendum.substack.com/p/are-llms-the-best-that...
Maybe those sunlit uplands are actually hallucinations? Maybe we're at peak-LLM?
Not sure if tech better = product better is a type of "technological determinism" (but it's certainly a fallacy) @jackstilgoe.bsky.social and @stsucl.bsky.social ?
The tech getting better does not mean that the product will. The incentive structures (slop, engagement maxing, attention economics) on the internet as it is are ominous.
I think today's LLMs might be the best we ever get
I helped write a thing (just in case you are not tired of reading about AI) open.substack.com/pub/asimovad...
Are LLM's the best that they will ever be? A guest post on Asimov Addendum by @rufusrock.bsky.social
"The technology might continue to get better, but that doesn’t mean that the user experience will.
Why? Primarily because no one knows how to make LLMs profitable."
Read more below👇
UK government simultaneously replaced expert advice on technology with tech company “consultants” and displaced their civil servants’ experience and expertise into AI models… It’s a disaster & an absolute gift to the tech vendors…
I wonder if he trusts plumbers, electricians, civil engineers, etc, who build and maintain all the infrastructure he needs to survive…
Public Sector Capabilities Index ➡️ www.ucl.ac.uk/bartlett/res...
Council on Urban Initiatives ➡️ www.ucl.ac.uk/bartlett/res... 3/3
To be fair, complexity is hard to work with. Humans sort of have to break problems down in order to structure them such that we can work on them. Very hard to sit down and write an entire book.
I like “technological solutionism” as discussed in Evgeny Morosov’s wonderfully titled “To Save Everything, Click Here”
Do you think that separation of state and establishment is likely shift ever? E.g., if certain places continue to feel neglected over time through different political regimes …
Interesting. So the establishment is not equated with the state? Makes sense through the lens of US libertarianism now I’m thinking about it. Harder to do in the UK with royals, Thatcher, etc.
I wonder why. Plenty of US cities/regions - like the Rust Belt - might rightfully feel aggrieved or left behind. The “coastal elites” vs everyone else narrative feels like a real thing too, yet national identity holds firm everywhere?
Are there parts of the US where people don’t strongly identify with “American”? Like how some in Liverpool reject being called “English”?
Another classic case of “let’s automate everything don’t worry it’ll be fine” www.bbc.co.uk/sport/tennis...
Great thoughts on information over abundance from @hankgreen.bsky.social — though calling a lot of the AI slop on TikTok “information” feels generous.
youtu.be/9euKCrTyMEc
When I worked in UK academia I recall hearing of a Japanese man who was baffled at how Britain had decided to run its universities like firms. “Why? Your universities are excellent and your firms are terrible.”
