The app for independent voices
For those of you who prefer a TL;DR:
I’m losing all my trust in the AI industry. It is facing many urgent problems that it’s not addressing adequately. The list I share here is not definitive but intended to spark conversation.
Disclaimer: Not everyone in the AI industry is like this. Also, I’m a supporter – I’d like to see it do well. That’s exactly why I’m writing this.
Talent churn reveals short AGI timelines are wish, not belief: When everyone is doing job hopping, it’s hard to believe their stated AGI timelines are truthful. You don’t jump ship in the middle of a conquest unless you don’t truly believe you’re conquering anything. Their stated timelines are “AGI is near” but their revealed timelines whisper “AGI may or may not happen at some indefinite point in the future.”
The focus on addictive products shows their moral compass is off: What’s with so many AI companies now working on making their products more and more engaging? I know this is to make money but come on, you have the most interesting and valuable technology in your hands and you resort to attention-hacking? AI girlfriends, flirty audio-bots, AI in short-form video, customized video games…
The economic engine keeping the industry alive is unsustainable: How can the AI industry survive when basically all standalone labs are money-losing operations dependent on constant infusions of VC capital? Last year, David Cahn at Sequoia said AI had to answer a $600 billion question. One year later, the question keeps getting bigger and bigger. No product-market fit and companies providing services below cost is what Tim O’Reilly called “AI’s Uber problem.”
They don’t know how to solve the hard problems of LLMs: hallucinations, unreliability, jailbreaking, prompt injections, fallibility – the same problems that existed five years ago exist today. None have been solved. On the contrary, hallucination rates have gone up lately. What’s going on? Do they just not care enough? Do they not know how to solve them? I don’t know what’s worse. The worst part is they keep talking about AGI and the alignment problem as if none of this mattered.
Their public messaging is chaotic and borders on manipulative: One day it is “AI will usher in an age of abundance, curing cancer and educating everyone,” and the next day it is “AI will destroy half of entry-level white jobs in five years.” Don’t they realize that panic paralyzes and unwarranted optimism feels like gaslighting? Some people at the top of the AI industry are so untrustworthy that they can’t possibly expect the world to just believe they will do the right thing when our fate is in their hands.
Andrej Karpathy: “You are getting way overexcited with AI agents.” One of the most well-known insiders is telling you: 2025 IS NOT the year of AI agents. The AI industry keeps hyping something that doesn’t exist yet! Like self-driving, it will take much more than it looks now. Karpathy says it’s the “decade” of agents. For now, human in the loop, keeping AI on the leash and partial autonomy are the ways to go.
