Grok3 Launch [video]

The man has almost unlimited wealth and his motivations seem consistently petty and strange or just downright ludicrous. He's like an alien to me.

I've noted the same feeling when seeing VCs/business people speak when I've encountered them.

LLMs are pretty good at translation between human languages which makes sense since they are language models after all. They are better at this any any other technology.

The state of the art image ones can also probably do OCR and handwriting recognition better than any other software though may be expensive to run in large volume. But if you need to take picture of a notebook page with your camera phone an LLM can quickly OCR it.

- Discussing the various stages of candymaking and their relation to the fundamental properties of sugar syrups, and which candies are crystalline vs amorphous. It turns out junior mints are fudge. Fondant is really just fudge. Everything is fudge, my god.

- Summarizing various SEC filings and related paperwork to understand the implications of an activist investor intervening in a company

- Discussing the relative film merits of the movie Labyrinth and other similar 80s kitsch movies. ChatGPT mentioned the phenomenon of "twin films" which was an interesting digression.

- Learning about various languages Tolkien invented and their ties to actual linguistics of natural languages and other conlangs

- Some dimensional analysis of volumes, specifically relating to things like "how many beans are in the jar" estimation and what the min and max of a particular weight of coins might be valued, in terms of both a par value based on a standard coin mix and outliers of, for example, old dimes that are pure silver.

- Discussion of quines in prolog and other languages, which resulted in a very interesting ChatGPT bug where it started recursing and broke when trying to write a prolog quine.

- Back of the envelope economic calculations around the magnitude of the housing deficit and the relative GDP cost for providing enough housing quickly enough to make an impact. Spoiler: it's probably unreasonably expensive to build enough houses to bring down housing prices by any significant degree, and even if we wanted to, there's not enough skilled workers.

- A number of podcasts transcribed. (I hate audio and meandering, so transcribed and summarized is perfect) I could use whisper and a python script to do this, but I'd rather let ChatGPT do the legwork, and it actually used a more modern model and method of processing than I would have naively used.

I find Github Copilot to be a really great autocomplete. I frequently write the comment at the top of a function and hit tab and it writes the whole function. This is dependent on typescript and having a relatively standard codebase but I think those things are useful on their own. You really have to limit it in terms of scope and specifics, but it lets me think high level instead of worrying about syntax.

Can we stop for a second and just marvel at a new piece of human ingenuity? Let's not give Elon too much credit, but I think that AI as a whole helps us all understand the nature of intelligence, and therefore humans' place in the universe.

One of the fundamental questions of human existence is: what does it mean to exist and think? Every time we build a new human-like thing it helps us understand the context of our own existence. (Not just computers or AI, but also airplanes, factories, etc.)

True AGI would force us to rethink what it means to be a thinking human being, and I think current LLMs already should and do.

Remember he was way less crazy before his market cap skyrocketed. Now he can afford being polarizing as a PR strategy once his fanbase has reached a certain critical mass. He’s been constantly testing what works.