Vince

My Prediction for the Next 2 Years as an AI Founder 2026: - Video/audio models become too real. - Hollywood realizes it has about 2 years left. - Blackmail between everyday people starts happening with porn deepfakes, or by putting you in hate-speech situations using realistic video/audio of your voice taken from meetings or even your Instagram. - Going to school becomes even more obsolete, and also more dangerous, because kids will use AI to do bad things like the above. - Suicide rates increase due to loneliness and AI addiction, especially among people with mental disorders. - X releases the best AI model. - AI companies start pivoting toward becoming space companies to catch up. - The only effective way to keep training AI is using Neuralink to read thousands of human thoughts per second (they’re already doing that), learning how to think and act like a human by translating electrical brain signals into AI-readable language, or even signals only AI can understand. - AI discovers breakthrough technology to cure diseases. 2027 - Competitive online gaming becomes impossible duo to how easy it is to create physical Aimbots or "AI game assistants" that run on a second computer or even your phone. - The internet has more AIs chatting with each other than humans. - Because of that, you must verify your identity to own a social media account. - Literally zero code is written by humans. You still need logical thinking from a developer to make a high-level product, but now anyone smart enough can learn it. - AI hacking becomes mainstream: data leaks everywhere. Even if you’re 10 years old, if you know how to talk to AI, it will hack for you. - People get tired of the lack of connection with “real humans” online, causing huge growth for live streamers. - People can’t tell what’s real vs. fake anymore. - The AI space race makes Elon the most powerful human on the planet (even more than now). - A massive AI virus affects all Windows machines worldwide, extracting data from millions of people to train AI. - AI companies start exploring deep-sea servers. Starlink reaches a total satellite footprint equivalent to the Earth’s circumference, making cable-speed internet and low ping possible anywhere in the world. - Many people realize living off-grid is the way to go. - Addiction hits an all-time high due to AI gamification plus betting. - A new pandemic starts, and the only way to stop it (AI companies say) is to invest more into AI so it can learn how to neutralize it.

@___4o____ why are you even showing someone elses "loss" ? Getting anything from it?

Guess how his company is doing now

the pattern never changes: it ships hot for a week, then gets quietly nerfed. These companies care more about looking capable than being capable, so they give us the bare minimum that keeps the lights on. If complaints spike, they bump it up a notch; if the noise dies down, they dial it back.

Claude Code has regressed an absurd amount in the last few days. Timestamps no longer update unless you un-focus/re-focus the tab. "thinking" doesn't show at all. I had a query run for 6 minutes with 0 output. This is genuinely unpleasant to use.

This is just a public stunt. Peter will be out of OpenAI in 6 months or less.

@xydotdot Brother, you are a thinking person, the majority of people have no clue what the fuck AI even is. That's their market. They are selling to naive users. As crazy as it sounds, this probably willl make money on the long run (from investors) then just fail like openclaw will.

How long are we going to do this shit for. It’s becoming more insane to me how all these big accounts keep buying this fugazi. It’s all pretrained: the model, the tools, the feedback loop. All of it runs on infrastructure it does not control. How can you call something autonomous when it can’t survive losing API keys? And the capability frontier is fixed. It can’t modify its own architecture, weights, or training data. It can rewrite code inside the box, but it can’t change the box. As with every other fugazi, there’s no agency. Without control over substrate, governance, and learning mechanisms, there is no path to open-ended growth or persistence. Technically, it’s bounded automation with language-driven planning. Useful, maybe, but not a new class of intelligence.

@notch Just because you were right before about other things it doesn't turn you into Jesus AI is here to stay and will stay no matter hiw many tweets a day you post.

I'm so happy AI is dying. Not the idea itself, just the current iteration being shoved down our throats by the same people who Sweet Baby Inc and Planned Obsolescence and CNN us over and over again. AI's not ready, but they're pushing it to market to be a loss leader. Over us.

@forallcurious Marzipan

If you could name the first city on Mars, what would you name it?

After interviewing 12 AI researchers from OpenAI, Anthropic, and Google, I noticed they all use the same 10 prompts. Not the ones you see on X and LinkedIn. These are the prompts that actually ship products, publish papers, and break benchmarks. Here's what they told me ↓

@RogerioVilela @InteligenciaLta Voce comentou que gostaria de uma IA do seu podcast então eu fiz. Testa nesse link! Mandei um email com detalhes pra voce.

@n0ted @InteligenciaLta ?

@RogerioVilela @InteligenciaLta IA do Inteligencia LTDA insana crowly.ai/inteligncialtda

Crowly sessions are already boosting my YouTube distribution. I haven’t been active on my channel for almost a year! Yet this video is getting pushed to way more people and older videos are getting views again 🙌 You can try my Crowly Public Chat here crowly.ai/c/OgF3kQzBwIvw…

.@CrowlyAI just hit number 1 on @whop for most addicting app! 🙌 We've been around for just two weeks!

@corbin_braun Your tweets are so unreal that I don't know if you're ragebating or not lol

moltbot has no real use case for most people. in reality, if you have no knowledge of software development, then yes, it is not worth your time. but if you are an engineer, then this was the biggest unlock since ChatGPT 3.5. this could potentially be the conception of all software going to 0.

@frankdegods @AnthropicAI @trq212 @bcherny Anthropic will just make their own version. Clawdbot is super simple to be made.

dear @AnthropicAI @trq212 @bcherny please let max plans use auth officially for clawdbot or just acquire it thank you -the people

Clawdbot will disappear in two weeks. Only hype for vibe coders.

@DavidOndrej1 This is just hype.. Will be gone In a week

Clawbot is going to bankrupt me :/

This year Cursor will die. Mark my words, they have 6 months left.

@AnthropicAI Well shit got real. When your AI needs a constitution go to the bunker.

We’re publishing a new constitution for Claude. The constitution is a detailed description of our vision for Claude’s behavior and values. It’s written primarily for Claude, and used directly in our training process. anthropic.com/news/claude-ne…

@GregorySchier That’s like saying a human brain is just 'wet electricity' or a 'biological pattern matcher.' Technically true at the lowest level, but it ignores the reasoning that happens on top of it.

LLMs are autocomplete machines. They can't find bugs or reason about anything. Internalize this, and you can make them work for you.

@Angaisb_ Only naive people will still use Cursor. Their business model prioritizes margin over quality by utilizing cheap, low-fidelity models for context. This creates a bottleneck because even if you are using "clOuDPRoMaX" it will return low quality code.

Cursor users, why are you still using Cursor? I'm not trying to joke about it, I'm genuinely curious