ialwayslikedgrime

we tend to overestimate how much others care about us and underestimate the impact and influence our actions can have on others

@mishushakov @DominiqueCAPaul @blwiertz but I agree on the skillset. the truth is: if cs people are mainly male, and then among the cs builders the female percentage is even lower, what can the poor guy do? like this guy is trying to make europe like SF and must also be the policies expert it's ridicolous

@mishushakov @DominiqueCAPaul @blwiertz as a F, I appreciate your input, but I personally disagree on the messaging part. The "come and build no sleeping" mentality resonates with me, and diluting hackathons' spirit wouldn't do anyone favors. If someone can't handle the intensity, then hackathons simply aren't for them

People often tell me there aren’t enough women at the events I run. That I should highlight them more, make it more visible. But when I check my YouTube analytics: 98.8% of viewers are men. Anyone can watch. No one is excluded. Yet, it’s almost entirely men. Why? Honestly, I don’t know. But I also think it shouldn’t, and doesn’t, matter for what I do.

@blwiertz as a woman who is interested in AI/tech/finance etc., I find it interesting how the algorithms think I am a man (suggesting red pill content, how to approach women - lol). Maybe women are free to watch your content, but it just doesn't reach them because... they are women.

the only skill that matters any more is the ability to attract attention. soon with ai, everyone will be able to do everything. knowing calculus, or anatomy will be entirely pointless. however, knowing how to draw attention to anything will always be valuable. if you do not understand/can't even imagine what the future is going to look like under an attention economy, you are genuinely fucked. attention will soon be all that matters, and it already is going that way. this doesn't mean you have to be an influencer, but if you know how to get attention, that's all that matters. you can sell litterally anything if it has a billion views, (i think roy lee said this) and getting a billion views is far easier than any of you would ever think.

i know there’s hugging face now, but what about the leaderboards… damn

i still need help to get over the fall of #paperswithcode. how am i supposed to easily identify the SOTA?

🏓🤖 Our humanoid robot can now rally over 100 consecutive shots against a human in real table tennis — fully autonomous, sub-second reaction, human-like strikes.

Photoshop is cooked, Nano Bananas manipulation is insane

companies like Facebook record every imaginable interaction their users have with the platform. they log each of your clicks and taps. they keep track of how long your gaze lingered on a post, whether you were on the same WiFi as that woman who might be your friend, which instagram reel you watched three times. for a single user this is quaint, but these practices are done on a planetary scale across all technology giants. they create petabytes of data per day and keep it for as long as the European regulators will let them. then they can have machine intelligence instrument it into useful knowledge for their cybernetic control systems that build newsfeeds, serve ads, decide how much compute to spend on you, which SKUs should be in which warehouses right before you want them. the Hive metastore bills run into the billions hospitals throw most of their data and telemetry out after each case, every single day. they record videos of vascular surgeries, endoscopies, discovering interesting physiologies. sometimes they're not recorded at all and most of them the time they delete them as soon as they’re done it's even worse for physiologic waveforms (ECG, EEG, arterial lines) which are essentially never recorded anywhere at all. milisecond scale views of patient's brains, vasculatures, hearts are generated and instantly destroyed. all of these time series of course predict people's hearts stopping, brains exploding, etc ahead of time. surgeons teleoperate robots, none of the micro-movements are recorded, policies never learned, never correlated into which outcomes were successful or not this would be unthinkable to most software people whose instinct is to record everything everywhere never mind the cloud costs, because we are sure there will be some use for it later and some model to be trained later. i don't have a prescription here per se my point is just that our civilization routinely hoards and treasures some of the silliest data in the world "i pressed like on the john pork reel" & destroys much of all the most important data it generates and limits what machines can learn

@gunsnrosesgirl3 whatever you do, that is not watching tv (today would be scrolling) is "useful". Pretty much the Steve Job's talk about connecting the dots.

What piece of advice from your parents turned out to be 100% true

I also suspect that positional encodings have a role in becoming able to imitate personal style. Each of us places commas and periods in a different way. Original transformers used sinusoidal and cosinusoidal functions. too regular and general imo

@emollick I think there could be emerging capabilities. With "emerging capabilities" I mean: text coherence; creativity in what's generated. In my opinion, the key is all those conditional probabilities of word after word of autoregressive models - cited this post on my profile

We really have not made a lot of progress on explaining the deep mystery of LLMs: How does a model using matrix multiplication to predict the next word manage to simulate human thought well enough to do all the very human-like things it does? And what does that mean about us?

It’s sad that a lot of modern architectures are not open to the public. Researchers have first to try to reverse-engineer how models are built. And then try to study their functioning. it’s a loss for humanity.

Maybe it is not as clearly staged as CNNs, but I find it to be as interesting as how CNNs typically detect low-level features like edges in the early layers, while deeper layers integrate them into shapes and objects.

I genuinely believe that there are emerging capabilities. With "emerging capabilities" I mean: text coherence; creativity in what's generated. All those conditional probabilities of word after word of autoregressive models.

We can now train AI inside the mind of another AI. 🤯 🌍 Our world model, Genie 3, imagines and generates new worlds on the fly. 🤖 Our embodied agent, Sima, is dropped in and learns to navigate them autonomously. The entire loop—from the environment to the action—is generated by AI. This is just the first step, soon we will have world simulators for training general embodied intelligence.

It doesn't give the full solution immediately. It gives hints and asks questions. It feels like coming to the solution together with little victories along the path. It feels like a very patient and experienced tutor. I love it! Thank you @AnthropicAI !

Claude is instilling this love in me again. I am coding on my own this time. I am committed to knowing 100% what I am doing, getting a refresh in coding from 0 to 100% on my little Genetic Algorithms side project. I provide the code that I write, and Claude tells me the mistakes.

Thanks to @claudeai 's new function "learning," I found again the love for learning how to code that I felt I had lost a bit since ChatGPT came out.