PhD of physics | Surviving humans behind this tech or Surviving AI? |

Joined February 2025
Pinned Tweet
What percentage of your dreams and fantasies have actually happened to you? Same applies to fantasies of a future with AI
8
Scam decoded: Open AI isn't going to seek a govt bailout but they do need the US govt & tax payer to pay the bill for the $1.8 trillion electric grid overhaul and data centre build required for their company to one day make a profit.
You tell a big story about a world changing innovation. Venture Capital floods in. Your inventions fails. You hide the evidence & hype harder for another funding round. Scared of sunk costs VC props up your lie. You hit the reality wall. Will VC learn after the AI bubble bursts?
I did a test with Grok recently - asking it to find quotation. It gave me one & claimed it came from a famous author. I checked it. The author never said that. I asked Grok to find the real source of the quotation. Grok attributed the quote to another author. I checked it. It was false. I did this 3 times and each time Grok gave a false attribution for its fake quotation. Run this test for yourself. If students & journalist use chatbots to find quotations, we're going to end up with mass false attribution & slop contamination in all our texts.
I'm helping students write essays & finding a lot of false quotations made by LLMs. We have to warn students not to use LLMs for essays, as 92% of students are now doing so & they don't realise they're putting false quotations into their work & fake history into our culture.
“There can be no assurance that the usage of AI will enhance our products or services or be beneficial to our business, including our efficiency or profitability,” Meta wrote in its 10k form last year. 🔗 on.ft.com/4oW1cDB
2
34
3
167
Anda retweeted
AI guy: *Posts picture* AI guy: Check out my amazing OCs and super original concept ^_^! Me: What was the prompt? Put StarFox characters in Swat Kats? Me: *Is blocked* Me: Lol. 🙄
5
2
60
Anda retweeted
what did Ilya *actually* see? scoop @theinformation: “a consistent pattern of lying”
The AI people keep saying this, but all you have to do is look around a little to see that it's absolutely not true. Far from it, actually.
This seems so so so dangerous. This is what has finally broken me and started turning me against AI as it currently stands - it's clear to me ethics don't matter to the big ai companies. This will fuck the people so hard - this will be an eventual economic bloodbath - am I missing something? How do you justify this
Guys, this is so important to understand fully: The market is being dragged up by only about 10 companies, all AI, all round-tripping revenue, all making outlandish claims of future earnings while producing none of it in real life. They are now actively coordinating public relations messages and cross-sells to each other in order to continue to push the bubble up. This is extraordinarily dangerous. I’ve not seen this before in my entire career. It’s a bubble that makes the dot-com era look like a pimple in comparison. Laying the current BS aside, and taking their claims at face value, investors need to realize that we will never get there. Not for a decade or more. We don’t have the energy infrastructure to provide for the data centers to even get remotely close to the earnings these companies have claimed. And if they do magically transform our society, somehow, what it will mean to you is far fewer available jobs and much, much higher energy costs. And yet, they drag the markets up every day. Eight out of 11 sectors, fully 80% of the companies in the index, were DOWN yesterday.
Like I keep saying, everything that comes from these things is made up based on existing patterns and they should be viewed as being for personal entertainment purposes only.
AI chatbots can be dangerous, but we also need to be careful not to anthropomorphize those dangers. excellent, nuanced thread:
1
7
Anda retweeted
e/acc doesn't realize they are doing Altman's marketing for him, just parroting these talking points.
3
1
6
Anda retweeted
the lesson of 2025 is that rich people can steal anything, and call it their own.
P1: So what is your job? P2: I wipe my buns on everyone's faces, including my paying customers! P1: Oh! You work in commercial AI!
1
11
Anda retweeted
Third party AI training dataset, AKA a treasure trove of copyright infringement that you think you're not responsible for profiting from because someone else did the dirty work.
Replying to @EpicArtFail
we're using third party!
5
62
1
568
Anda retweeted
Sneaky Zuckerberg: hire these folks by offering them billions, ask all the questions about how the competitors do things, then cut their positions👍
Anda retweeted
I'm so sick of you constantly talking about how AI hasn't taken jobs because you're operating on this technicality that it has to literally do everything a person does, rather than just replace 80% of the work.
Replying to @wittycarbon
AI hasn't taken any jobs. No one has ever hired an AI to do exactly the same job for which a person was previously hired. AI has been and will always be a tool. Today, this tool is more versatile than before, but it's still a tool, and it takes an engineer to use it.
9
58
464
Anda retweeted
People don't hate this man nearly enough.
Wait… so users of OpenAI’s Atlas browser can opt-in the web pages they browse - *which belong to other people* - to AI training? Cool cool openai.com/index/introducing…
Anda retweeted
They're trying to corner the market on literally everything and destroying the internet to do it.
Meet our new browser—ChatGPT Atlas. Available today on macOS: chatgpt.com/atlas
Anda retweeted
i will not be installing ChatGPT Atlas browser from OpenAI and neither should you