ma2rten

This is not correct, I am a co-author of the chain of thought paper and we reported GPT-3 results.

The checkout lane at the grocery store is not really the pinnacle of meaningful human interaction.

During the Microsoft presentation they said "This is a new built from the ground up with search in mind" or something to that effect. But it's unclear what that means exactly.

At Google it is actually common to have directors or VPs who don't have reports for a while, but we are talking about people who are ICs already. For each management position there is an equivalent IC position at Google with similar pay. I believe L9 would be a senior director.

I'm the opposite. I was using my laptop's build-in screen for months. My manager had to remind at every 1:1 for months to buy a monitor until I finally did.

You could probably do much better than ChatGPT if you built a bot specifically for being a lawyer.

I'm pretty sure if you'd just prompt GPT-3 it would give an appropriate response to this. Less than half on hour of work.

Why does the title and beginning of the article talk about ChatGPT if he actually used InstructGPT?

As someone working in the space, it's quite strange to observe how ChatGPT blew up, when OpenAI had similar models for quite a while.

This makes no sense, you benefit from the low prices if you are not buying.

a) It's only about 6%

b) Google grew by 30k last year.

c) Google is such a profitable company, because their revenue is not directly tied to number of employees.

1. It doesn't contextualize that $2 is a good wage for Kenya.

2. It says OpenAI "used" workers which has a negative connotation.

3. It is factually incorrect: OpenAI paid a vendor $12.50 per hour. It's unclear if they knew how much the actual workers were paid.

4. It used a negative framing implying ChatGPT is toxic.

So something like

"OpenAI vendor paid Kenyan workers above median wage to make ChatGPT more safe"

It's not public yet.

No, you need billions of words to train a large language model.

I have access to Claude. I think the length of the responses vary more depending on the context than at least the previous version of ChatGPT. It would give sometimes longer, but also sometimes shorter answers and it would feel better overall. But ChatGPT recently had an update and it also improved, so now I'm not sure anymore who is better.

Ah, I misread the bullet points as pointing out problems.

This is news to me. Do you have more information about that?

This post is about image generation, not language models.

DreamBooth works with just 4 input images.

If that's what they want to work on?

And almost none of the commenters seem to have read the article, so everyone is saying how predictable this was. Quite bizarre.

We will see an explosion of AI capabilities in the next couple of years. This will have a huge impact on our lives, much of it good but some of it also bad.

Did these podcasts have transcripts? You might be inadvertently evaluating it on data that it was trained on, which is basically cheating. Even if not, it might be trained on similar podcasts. Judging how good these kinds of models are is really hard.

I'm looking forward to your comparison. It's really hard to make sense of how good this model actually is without being an expert in the area.

A lot of this is because large companies which make proprietary software, such as Google, invested in the software you listed. I am not sure if the same would happen in the case of hardware. There is more upfront investment. The culture is different, more secretive.

Those images are not really "labeled". They just scraped the alt text. A lot of the recent advances in AI have been by using lower quality large scale web data, instead of hand labeling. The noise will average out. Hand labeled data can be used for finetuning.

Going vegan and eating different types of eat are not the only options, many people these days choose to eat less meat for health reasons.

Yes, check out Neural Radiance Fields (NERF).

This doesn’t make sense to me. Why wouldn’t just the 1x programmers unionize? I’m sure they would be in the majority as 10x programmers are rare.

I also do not really believe in this concept in general. Programming skills are not one dimensional, people have different strengths and weaknesses.

It might help in this situation, but it's not the best possible solution. Storing it in USD or gold is more stable. You could keep those in a vault or in a foreign bank account. Bitcoin is very volatile and, as you said, still requires exchange, which might also be very tricky in a situation like this.

Your description sounds like he robbed the bank to me.