Rufus Rock

Are LLMs the Best That They Will Ever Be? Without careful disclosure architectures, AI products may only get worse, even as the technology gets better.

@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...

Are LLMs the Best That They Will Ever Be? Without careful disclosure architectures, AI products may only get worse, even as the technology gets better.

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/ phttps://asimovaddendum.substack.com/p/are-llms-the-best-that-they-will?r=3f5ape&triedRedirect=true

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

Are LLMs the Best That They Will Ever Be? Without careful disclosure architectures, AI products may only get worse, even as the technology gets better.

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 IIPP, with support from Bloomberg Philanthropies, has launched an initiative to establish a Public Sector Capabilities Index – the first global measure of city governments’ capacity to solve problems.

Public Sector Capabilities Index ➡️ www.ucl.ac.uk/bartlett/res...
Council on Urban Initiatives ➡️ www.ucl.ac.uk/bartlett/res... 3/3

Hope, Climate Change and Technological Solutionism On the 20th of September, 2023, Rishi Sunak delivered an update on the UK’s approach to Net Zero, bringing technological solutions once again to the forefront of climate policy. The “new approach”…

I wrote a thing about this in the context of climate change a while ago medium.com/@rufus.p.roc...

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”?

Wimbledon 2025: Play stopped in Sonay Kartal's match against Anastasia Pavlyuchenkova after electronic line calling fails Play is stopped in Sonay Kartal's last-16 match at Wimbledon as the electronic line calling system fails to call a ball out.

Another classic case of “let’s automate everything don’t worry it’ll be fine” www.bbc.co.uk/sport/tennis...

YouTube video by vlogbrothers You’re Not Addicted to Content, You’re Starving for Information

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.”

Europe’s dependence on US tech is a critical weakness | Opinion In order to protect their sovereignty, the continent’s leaders must invest in a digital ecosystem independent of America

Europe’s dependence on US tech is a critical weakness | Opinion

https://www.ft.com/content/30d6f79f-d1ee-49dc-bff5-719f18c1a9e5

Open letter to the President of the Royal Society – time to stand up for your values If you wish to show your support for the letter below regarding apparent inaction by the Royal Society in the face of breaches of its code of conduct by Elon Musk FRS, please sign below. I invite anyo...

Signature tally now closing in on 3000. Please keep discussing, sharing and signing to put as much pressure on the Royal Society to respond as possible. forms.gle/AuYvuWqytx8K...

The End of Programming as We Know It

I just wrote a piece on why I think the chatter about AI putting programmers out of work is wrong. Every time it gets easier to tell computers what we want them to do, it brings more people into the fold and unleashes a wave of creativity and creative destruction. That reads like opportunity to me.

'Scaling laws' are not laws. They are an industrial strategy. Like Moore's Law. Peter Thiel, Reid Hoffman et al's airport books lay it out - Blitzscaling as a way to crush competition and earn monopoly rents. So the story has to be that bigger is better.