Dr_Birdbrain

When I was a little kid, my toddler sister fell into a pool I was in, and this was her exact reaction, she just froze up. A lot of family was nearby and one of my uncles scooped her out, but the freezing up is such a surprising response, and apparently not uncommon.

I bet they are training their internal models on the data. Bet the real reason they are not training open source models on that data is because of fears of knowledge distillation, somebody else could distill LLaMa into other models. Once the data is in one AI, it can be in any AIs. This problem is of course exacerbated by open source models, but even closed models are not immune, as the Alpaca paper showed.

I think assembly theory is most useful to define a spectrum of complexity.

Sure, a refrigerator requires intelligence, whereas loose carbon atoms do not.

What about everything in between? Methane? Complex hydrocarbons? Pointy sticks? Things that look like crude tools?

It’s about defining a metric that can measure complexity across a wide range of things you might encounter, to help estimate the odds that it was produced by something simple yet alive, or something complex and alive, or something alive but not very smart, or something kinda smart, or something super smart, etc

My thinking exactly.

Would that help at this point? Or has the culture fundamentally been changed?

I think the raw text inside Facebook groups is at least as valuable as Reddit data. Even if demographics data is restricted under European law, the raw text of people interacting is quite valuable.

Read paragraph 2: the best counter argument is the classics themselves.

Skimming through it I was confused.

- No homegrown semiconductor industry: isn’t the recent hand-wringing over the new Huawei chips proof of the opposite?

- No interest in training LLMs? Is that true? I thought Baidu was already on it?

In fact at every major AI conferences, Chinese R&D groups like Baidu and Ant group are major participants (and sponsors). I am talking about conferences like NeurIPS and AAAI, which both happened in the past few months.

EDIT: the comments of that article are also confused by that article, lol. Is there a joke that is going over our collective heads?

Do you add any oil?

Yes! I like Paganini’s Caprice No. 24

Pretty sure he’s leaving BILLIONS on the table :)

To dedicate himself fully to his YouTube channel? I am looking forward! His content is amazing

What is the expected dealmaking rate for these folks? Her manager did not express dissatisfaction with her performance. She said she was ramping up for three months and properly working for one month, how many deals are they supposed to make in a month?

I’m not in sales so I have no idea about the expected dealmaking rate, but if she really were subpar, the people doing the firing would have had a much easier job. They wouldn’t have hemmed and hawed and said they couldn’t discuss particulars, etc, they would have said exactly what you said, “by your own admission you failed to close any deals, the expected dealmaking rate is X many deals, etc”

Or, give a tax deduction for contributions to open source.

This is America, new taxes are nearly impossible to pass, tax cuts are much easier to pass.

Nah, if it ever starts to look like NYT might win, MSFT will just do a hostile takeover.

Market cap of NYTimes is 8B, OpenAI is 80B, MSFT is 2.7T, you do the math.

I have a question about this—-isn’t it against the OpenAI Terms of Service to do this?

Are we sure this is true? The last few years we have seen huge amounts of psychology research be discredited and overturned. Is it possible that the foundational ideas of advertising wouldn’t replicate, but everybody is too scared to look?

This is my reaction exactly—i can’t believe anybody is replacing humans with AI, given the current state of development of AI.

As an American, corporate health plans are excellent, even covering things like IVF. Pretty sure the bare-bones plans are not as comfortable, so yes, “I need to keep this job to have this excellent health insurance for my family”.

Yep it’s an incredible technique. I use it to remember key ideas from audiobooks, long enough to jot them down, sometimes the next day. Incredibly powerful

I also discovered outdoor calisthenics over the pandemic. Massive upgrade in every way :)

Yeah the signup can be rough. I remember they insist on offering you a “free session” with a trainer—it’s a trap, one must adamantly decline it, insist on just wanting a basic subscription. They’ll want to ask you lots of questions to “try to understand your goals”, one must insist on just a basic subscription.

It’s quite obnoxious :P but, that’s the only time one interacts with sales staff or even trainers. After that it’s pretty smooth sailing.

The ones I have tried were mild, very creamy, like a light and mildly fruity yogurt.

I’m going to say “hard”. That’s a beautiful piece of prose, both because of the social skills needed to elicit the behavior from Huang and because of the narrative skill to tell the story well.

Construction is expensive because it requires labor, and labor is expensive.

How much creatine, if I may ask?

XD I love this theory

I don’t buy it. “Move fast and break things” is the mantra of Silicon Valley. A buggy rollout is classic, and IMHO no big deal.

My best guess is she’s hoping to “divide and conquer”.

Q1 of 2024, unironically