Notices by Baldur Bjarnason (baldur@toot.cafe), page 3

“Microsoft’s Hypocrisy on AI - The Atlantic”

https://www.theatlantic.com/technology/archive/2024/09/microsoft-ai-oil-contracts/679804/?gift=lhL3dXSYCcu9vqTqEbg0OHfJiu_TRdq079IHN4QaSAE

> Microsoft is reportedly planning a $100 billion supercomputer to support the next generations of OpenAI’s technologies; it could require as much energy annually as 4 million American homes. Abandoning all of this would be like the U.S. outlawing cars after designing its entire highway system around them.

Fuck every single person who is helping this disaster happen

Since every new release by OpenAI is usually followed by nonsense commentary and bubble inflation: a note

Testing a large statistical model with a handful of manual tests ("it solved this riddle!") is like assuming the home remedy you took cured your cold. These are methods that only lead to superstition.

People who use these tools today are enjoying a short-lived grace period. It’s going to get much worse. Think general information retrieval is bad? Just wait until bad actors figure out how to better manipulate code LLMs

The problem with using tools that can't be secured is that the bad actors will ALWAYS win in the end. There isn’t a gmail endgame here where Google can close off the ecosystem to its own advantage. Information retrieval will always have to retreive information at some point

“Shady Firms Say They're Already Manipulating Chatbots to Say Nice Things About Their Clients”

https://futurism.com/shady-firms-manipulating-chatbots

> all it took were a few human-illegible text sequences crafted to manipulate AI training data, which AI researchers simply fed into a chatbot as you would any prompt, along with a simple plea on Roose's personal website imploring AI chatbots to say nice things about him.

This is very plausible as prompts effectively can’t be secured

RAG, which was supposed to fix hallucinations, essentially requires feeding query results directly into the LLM

Like some of us have been saying for a while now, these systems are simply not suitable for purpose and are inherently flawed. Worse yet, they are fundamentally insecure and prone to manipulation

If you think spam's a problem now, imagine a world where every spammer has sophisticated tools to control how THIRD PARTY chatbots talk about any given product.

The LLM honeymoon phase is about to end: https://www.baldurbjarnason.com/2024/the-llm-honeymoon-phase/

“Going Buildless | Max Böck”

https://mxb.dev/blog/buildless/

I've gone buildless for personal projects for a long while now and am not going back

Also on HTML includes: it makes me sad to think that XML had standardised includes almost 20 years ago. I wasn't a fan of XML at the time but in hindsight it's looking like we, web devs, dumped an entire ecosystem of tools and standards just out of an annoyance with error handling that we then got a worse version of anyway with JSX, setting us back a decade

So, what I’m taking from this “Founder Mode” discourse—without reading the original post, because fuck Paul Graham—is that tech companies generally have awful management, because tech execs are themselves awful, because the companies are founded by assholes who are incapable of managing others or hiring people who are good at managing others?

That in the ballpark?

The way AI fans keep having to say “AI is not going away” makes me think there’s absolutely a chance it might go away.

“AI worse than humans at summarising information, trial finds”

https://www.crikey.com.au/2024/09/03/ai-worse-summarising-information-humans-government-trial/

> The reviewers’ overall feedback was that they felt AI summaries may be counterproductive and create further work because of the need to fact-check and refer to original submissions which communicated the message better and more concisely. 

I'm shocked, shocked, etc etc

“Watch what they do, not what they say | LinkedIn”

https://www.linkedin.com/pulse/watch-what-do-say-glen-peters-skadf/

> If you are serious about your climate targets and net zero pledges, then you must do something different to what you did last year, and the year before.

“Is AI a Silver Bullet? — Ian Cooper - Staccato Signals”

https://ian-cooper.writeas.com/is-ai-a-silver-bullet

> But the flaw in prompts as a description of the code to be generated tends to be that natural language is a remarkably poor medium for expressing the model that we want to use in our software. Every time we move away from a 3GL to create a more natural interface, we rapidly get diminishing ability to author software.

This entire essay is excellent

I’m firmly in the not camp as software dev has a long long history of massive self-delusion and are composed of exactly the kind of demographic that are likely to fall for these kinds of biases: educated and supremely confident about their own intelligence. There’s also a lot of money in LLMs, which means there’s a lot of pressure to play it up

Software is also largely managed by outright sociopaths

So, I generally don’t trust anecdotal evidence coming out of the software industry.

I will note that I’m much more sceptical than the author is of copilots. Most of the studies I’ve seen point to a very high error rate (security flaws, bugs, hallucinating APIs) in copilots, so the question is whether dev self-reports are accurate (studies lag reality) or not (devs are falling for confirmation and automation biases).

@inthehands Oh, I see. Yeah, those are excellent questions that are well worth digging into. I have similar doubts overall.

@inthehands It’s still purely anecdotal on my part, based on conversations and observations. It’s still to early to be sure, but I think that for some people, LLMs are fulfilling demand that OSS used to. The question is just how big that group is.

“The slow evaporation of the free/open source surplus”

https://www.baldurbjarnason.com/2024/the-slow-evaporation-of-the-foss-surplus/

Where I try to explain, as succinctly as I can (which isn’t that succinct), why I’m worried about where FOSS is heading

“Just use fucking paper, man - Andy Bell”

https://andy-bell.co.uk/just-fucking-use-paper-man/

This. Index cards too.

(I alternate between pencil and fountain pen depending on my mood, but the principle is the same in both cases.)

We don't have that much excess power generation. We even have to occasionally limit power delivery to heavy users

Increasing power generation that much requires tough choices: we'd have to ruin the environment some way. We just don't have that many locations left for hydro or geothermal. Most locations that remain are popular tourist sites—destroying them would be bad for the economy—important ecosystems, or would require improvements to the grid that nobody seems to be willing to pay for

Self-indulgent rant:

One reason why IMO, the case against "AI" is as cut and dried as the case against cryptocoins is that shutting those two sectors down through government action would immediately solve the household energy transition for Iceland, for example

The maths is fairly straightforward: switching personal transportation over to electricity here would roughly double household power consumption (give or take, with the assumption that current plans to improve public transport pan out)