Seth

@Leo Gura What should a company maximize if not profit? Isn't it the citizens' job to vote for politicians that will properly regulate the company? People just use CEOs as a scapegoat when they should actually be upset at themself or their neighbor for voting in the wrong politicians. It isn't a company's job to regulate itself. The regulation is the government's job. As long as they aren't breaking any laws, they are playing their role perfectly. Capitalism works. Correction: laws* not jobs.

@LSD-Rumi From Wikipedia: "Debate is a process that involves formal discourse, discussion, and oral addresses on a particular topic or collection of topics, often with a moderator and an audience."

Attacking a cop pointing a gun at you is a good way to get shot. The cop was fully justified. This is absurd. He wouldn't have shot her if she didn't try to throw boiling water at him. If you think boiling water isn't a threat then try sticking your finger in boiling water one time.

@Leo Gura Are you anti-debate? It seems like a crazy position to me. How can you be confident you know something if you avoid debates? The point of a debate is to see perspectives and arguments you haven't thought of. It is like learning surgery on your own with no one to point out your mistakes. It is like writing mathematical proofs and getting no feedback. It is like doing a scientific study, writing a report on it, and then never getting peer review. You can't be confident if you haven't tested a position in the intellectual space. This isn't about absolute truth, this is relative truth.

I would much rather see you speak with Destiny about the Israel / Palestine conflict. I feel like a conversation with Jordan Peterson wouldn't go anywhere interesting.

I used to eat a lot of beans everyday, one or even two cans daily. They taste pretty great, but I started feeling lethargy in my body and mind to the degree I would call depression. It took me a long time to realize the beans were what was causing it and I suffered a lot because of it. It is crazy that something as seemingly benign as beans can ruin a person. I was probably eating too many beans and I may be sensitive to them compared to the average person. I no longer eat a can of beans everyday. lol

The title of this post is somewhat misleading. As Dwarkesh Patel notes, Francois Chollet does not seem to disagree with the popular view that LLM intuition may be what drives a logical planning algorithm in future AGI. The primary disagreement appears to be about timelines rather than the required structure. Some have hope that reasoning will be solved this decade, while Mr. Chollet considers it a significant challenge that won't be resolved soon. Robert Miles has a great video explaining why a system 2 algorithm is so exciting for AGI. (I also recommend his other videos. He is a good resource for learning AI safety concepts.) With system 2, planning might not just amplify intuition; the amplified intuition could be distilled back into the neural network, improving the base intuition. Of course, the real world is more complex than a finite board game, and researchers are still seeking ways to make this process compatible with LLMs.

@Leo Gura Can you elaborate more about applying the logic of death ratios both ways? It sounds like you are conflating terrorism with collateral damage in war, but I have faith you wouldn't do that. I am confused.

Although not everyone agrees about the risks, safety is a huge issue in the AI community. It is talked about constantly. People also constantly complain about how much companies censor their models for safety (and mainstream appeal). It feels like you are assuming every billionaire or adjacent is evil now. What makes you so confident?

You're being sarcastic right?

David Shapiro gives me conman vibes. He quit his job and now needs to hype up AI as much as possible if he is relying on his youtube channel now. I hope I'm wrong and I hope David is correct about his AGI predictions.

You misunderstand that not all difficulty is good...

Your best bet with free math education is to find and join a mathematics discord server then ask them for book recommendations. You can also ask for help on problems you get stuck on and to check your work.

There is a lot of impractical advice on creativity out there. Here is a simple process I find effective: 1) Collect a diverse set of ideas (think 'requisite variety'). 2) Choose a main idea, either your favorite or a random one. 3) Select another idea and combine it with your main idea. That's it! Repeat this process multiple times, avoiding harsh judgment before putting ideas into practice. Intuition can be helpful here, but beware of being overly judgmental. The finer details of this process vary depending on your creative medium. Whether you have the mental energy to be creative is a whole other thing. If you find yourself feeling down after a meal or stressed and overwhelmed, taking a short nap can often be helpful. Assuming it aligns with your schedule, try setting a 30-minute timer and simply lay down. Remember, you don't have to fall asleep for this to work.

I feel like the biggest question, whether they are real or not, is that if they are real, then why hide?

How can liberals be more developed than conservatives when studies have shown that liberals are more mentally ill and less happy than conservatives? This seems to contradict the videos Leo recently made on these topics. How can one reconcile this?

A man that feels no fear. Check out the documentary "Free Solo". https://nautil.us/the-strange-brain-of-the-worlds-greatest-solo-climber-236051/

Imagine if you wanted to juggle but instead of using your own hands you tell a robot to do it for you.

Imagine if you wanted to go on a hike but instead of using your own legs you tell a robot to do it for you.

@Thought Art Try this one. It sounds very natural. Voice Actors are in trouble. https://beta.elevenlabs.io/s https://www.whyp.it/tracks/67397/copypasta?token=7gByh

@Carl-Richard "make a program distinguish between relevant and irrelevant information for performing a task" This is what transformers do.

Large Language Models like chatGPT are good at intuitive thinking (continuous) and bad at logical thinking (discrete) because of the way they are trained. I think need both to be creative and have truly original ideas.

One of my favorite instances so far is where someone told ChatGPT to imagine a virtual machine.

I personally feel like the quality of this forum has decreased in recent years.