Hrishikesh Hardikar

👨‍🔬🍛 🧬 🧪 🚲

[Craig Weiss @craigweiss]

sf nightlife is so back. went to a warehouse rave last night and there was a guy in the corner deploying to prod. everyone gathered around to watch

[Alexander @toofawlty]

I’m sold on AI JUST because it makes frontend development actually enjoyable.

I can finally come up with a design I don’t dread having to implement.

[Nassim Nicholas Taleb @nntaleb]

This was already expressed by Seneca who said that if you are trying to distinguish yourself with your strength, an ox will always be stronger than you. So work on being maximally human.

[Floro S. @sflorimm]

95% of code was slop before AI, too.

It was written by underpaid juniors because businesses wanted senior-level results on a junior-level budget.

[Hrishikesh Hardikar @HPHardikar]

@flxrnc Funny, when I asked it to integrate Clawd into an artifact I was trying to make, it refused saying it’s an official mascot of Anthropic and it therefore cant animate it.

[Hrishikesh Hardikar @HPHardikar]

And I’ll say kudos to Anthropic for sharing all these controversial reports. Would you prefer they never tell us what could go wrong?

[John Ennis @johnennis]

There are so many contradictions with Anthropic

They talk about ethics and say that their models might be conscious, but then they are happy to sell you the use of those models for $200 a month

They say AI is dangerous, but they are pushing the limits harder than anyone

They report out the unpredictable behavior of their models, but they want people to turn over a large batches of important work to their models

Yes, they make the best models, but there is something fundamentally unstable about the company

[cromwellian @cromwellian]

@genzzztrade Sounds like Sacks is just re-inventing Asimov's laws, and we know what the issues were. There's a reason why our law books are voluminous and not just a simple list, because the edge cases are where all the trouble is.

[Walt @u_nonnie]

He must be one of those black and white dichotomous thinkers. That does work to a point--until it's insufficient and you need the dialectical thinker. I'd say this demonstrates a deadpan ignorance, that, if he understood what he was revealing about his own cognitive limitations would justifiably be embarrassing. Claude can do dichotomous thinking already. What will make it smarter is the thing this dude calls "overcomplicated".

[Hrishikesh Hardikar @HPHardikar]

@TonyMcDow @genzzztrade And I also disagree with “hasn’t produced anything of value”. If this was a critique of GPT3 era models I would agree. 2026 models in general have been vastly valuable.

[Hrishikesh Hardikar @HPHardikar]

Asking with respect: pause and do what? Even if the government did decide to pause it, it would be a lazy way out. Do better. Make better rules. Hold people accountable. But, pausing instead of doing all of this is weird.

[Bernie Sanders @BernieSanders]

If you or I called in a fake homicide tip, fraudulently applied for 20 visas and hacked websites, we’d be arrested and prosecuted.

The same standard must apply to AI CEOs.

Prosecute CEOs when their products break the law and pause advanced AI NOW.

[float @notesbyfloat]

A man cannot prompt the same agent twice, for it is has compacted and is not the same agent, and he is not the same man

[Charles C. W. Cooke @charlescwcooke]

A lot of people are mocking this, but I think they’re wrong. I’m not kind to AI for the AI’s sake; I’m kind to AI for my own sake. It’s not good for *me* to be cruel or abusive.

[Hrishikesh Hardikar @HPHardikar]

@film_girl No it’s not. Bullying a chatbot is insane. Bullying anyone or anything is.

[Hrishikesh Hardikar @HPHardikar]

@wongmjane I completely agree with this. Living thing or not, human or not, bullying is unacceptable.

[Sebastian David Lees @sebs_tweets]

One thing we're learning with AI: If you make something 10% easier to use, people don't just use it 10% more; they use it in entirely new ways.

[Hrishikesh Hardikar @HPHardikar]

I am genuinely confused about different models @claudeai makes. Let’s take three:
Opus, Sonnet, and Haiku.

I understand “Opus for initiation/changes, Sonnet for edits”.

Where does Haiku come in? I don’t think its very useful for research work… or am i missing something?

[Hrishikesh Hardikar @HPHardikar]

Decision I do not regret: learning R instead of excel

[CG @cgtwts]

All those years of refusing to learn excel and PowerPoint is about to pay off

[CG @cgtwts]

All those years of refusing to learn excel and PowerPoint is about to pay off

[Hrishikesh Hardikar @HPHardikar]

@mauriciojornal_ I get you. I won’t say it’s “worth” it. YouTube makes you experience insufferable, and makes you pay to make it usable. Thats what you do when you have absolute monopoly. Other brands give you a taste of what’s possible and paywall the best experience. That’s what I call “worth”.

[Hrishikesh Hardikar @HPHardikar]

@bcherny I’ve always talked to it like I would talk to myself. That means sometimes my prompts contain chains of thought that I discard later down in the same prompt.

It works best for me. I’ve tried “do A then B then C” type prompts, not for me.