Destruction and Distortion of Information, Including Facts About Linux (Bonus: This is Destroying the Planet)
"You can put anything into an LLM and get back perspicacious analysis, right? So DOGE is feeding sensitive data from across the Education Department, including personal information about student loans, into a Microsoft-hosted LLM to work out what they can cut." -Pivot to AI, 2 days ago
LLM slop is a real problem. Unless people are berated for misusing LLMss where they don't belong (i.e. used as anything but a mere toy), society will be worse off.
All that LLMs have going for them is hype, and moreover media that intentionally misrepresents them and their supposed capabilities.
RMS and other people use the word "BS" (or "bullshitting") to explain what LLMs are. Two days ago this short article said "ChatGPT isn't hallucinating — it's bullshitting" (or worse). This is what a Bill Gates-bribed British publisher said about another Bill Gates-bribed British publisher. It was published yesterday morning:
The researchers asked the four generative AI tools to answer 100 questions using BBC articles as a source. The answers were then rated by BBC journalists who specialise in the relevant subject areas.
About a fifth of the answers introduced factual errors on numbers, dates or statements; 13% of quotes sourced to the BBC were either altered or did not exist in the articles cited.
Imagine society relying on LLMs for "news". Well, for some people that's already a reality, yet some fail to recognise that they read falsehoods spewed out by machines, hence they're not "lies", merely "hallucinations" (same thing in practice; the net effect is the same, sans accountability and legal questions).
Yesterday we saw this fake 'article' about FUD from the Microsoft staff and its parrots, some of whom simply misuse LLMs to amplify Microsoft staff inside the Linux Foundation (LF). This example is new:
It's LLM slop:
OpenSSF runs for (and was run by) Microsoft et al. It does not exist to establish real security but to lobby for dominant "LF stakeholders" (i.e. back doors or fake security, which is essentially the same thing).
Disinformation aside, much could be said about environmental aspects, which lessen access to food (energy bills rise, but food prices rise even faster). LLMs and other bloat are an attack on the only planet we have. As we've just mentioned Gemini Protocol (not the 'fake' Gemini), consider how much would be saved by replacing so-called 'webapps' with "The Small Web". "Social control media platforms became walled gardens," ['webapps'] argues this blogger. "Websites became bloated with analytics and ads. Loading a single page now means downloading megabytes of data. We traded simplicity for convenience. We traded ownership for reach. We traded independence for engagement metrics."
Not all of us. There's growing resistance to the waste and seeing there's plenty of wasted energy in the UK, I've witnessed firsthand people who rethink their power consumption for pragmatic reasons.
"The data center craze," one person told me yesterday, "where all municipalities are scrambling to build large, "300MW" data centers, is all about surveillance."
"If the centers were stocked with Arm, MIPS, or POWER, it would be otherwise. But using Wintel gives away the goal: surveillance, along with wasted electricity," the person said. There's also concern for water. This week the Southeast Asian media argues "Malaysia water regulator to set strict water rules for data centres as number grows" (waste of water where it could instead be used to grow food, not enable operations that poison the air).
"The guidelines call for the use of alternative sources instead of drinking water for their operations," it says. Alternatively they could cut down the number of data centres. Why is that not on the agenda?
Quoting The Register about this country: "While the UK government wants to turbocharge datacenter construction, a newly published report says there are already 400 GW worth of outstanding requests for connection to the power grid around London, and regulator Ofgem estimates 60-70 percent of these will never happen."
Who is served by this? The "Cryptocurrency Ponzi Scheme"? The "genAI Ponzi Scheme"? What about the rest of us? Who's gonna pay for these Ponzi schemes? By their very nature, these schemes pass all the costs elsewhere.
Society stands to gain nothing but misinformation from chatbots (which is what LLMs typically are). And chatbots are behavioural control at scale, including the censorship, so they exist to annoy ordinary people, for example offering them fake "customer support". The people who run our government don't seem to mind British people talking to GAFAM instead of peers, whom they can collaborate with. In the corporate media they call it "arms race" because they hope hype alone can make it a "thing" (and then weaponise it; that's why there's growing fear of DeepSeek). See, when Western social control media like Facebook was gaining traction nobody panicked; the modality was worrisome only the moment China was in charge (TikTok). As social control media is still (mostly) not bots and those networks are manned by people rather than LLMs those are harder to control except with bans, deranking etc. Well, the algorithms are then becoming 'the arms race' (how to meddle in what people say and do so at scale, algorithmically, e.g. to suppress discussion of some topics).
"It already is weaponized," an associate said about social control media, "they are merely trying to scale it and expand the victim pool."
They also try to automate the biases and in the process they burn a lot of energy. For ordinary people it means higher prices (bills people can barely pay).
LLMs are a terrible thing no matter which aspect one considers. We need to banish them. Abandon sites or companies that use them. █