More sympathy for squirrels than human beings

Anthropic released an api for the same thing last week.

Every credible wiki has moved away from fandom at this point. All that’s left is the abandoned shells of the former wikis they refuse to delete and kids who don’t know better.

We’re Costco guys

This is actually pretty smart because it switches the context of the action. Most intermediate users avoid clicking random executables by instinct but this is different enough that it doesn’t immediately trigger that association and response.

All signs point to this being a finetune of gpt4o with additional chain of thought steps before the final answer. It has exactly the same pitfalls as the existing model (9.11>9.8 tokenization error, failing simple riddles, being unable to assert that the user is wrong, etc.). It’s still a transformer and it’s still next token prediction. They hide the thought steps to mask this fact and to prevent others from benefiting from all of the finetuning data they paid for.

The role of biodegradable materials in the next generation of Saw traps

It’s cool but it’s more or less just a party trick.

How many times is this same article going to be written? Model collapse from synthetic data is not a concern at any scale when human data is in the mix. We have entire series of models now trained with mostly synthetic data: https://huggingface.co/docs/transformers/main/model_doc/phi3. When using entirely unassisted outputs error accumulates with each generation but this isn’t a concern in any real scenarios.

Based on the pricing they’re probably betting most users won’t use it. The cheapest api pricing for flux dev is 40 images per dollar, or about 10 images a day spending $8 a month. With pro they would get half that. This is before considering the cost of the language model.

Most of this seems true (or was at the time) but this is outdated now. Mr. Beast is no longer managed by Night Media.

This is why you should always selfhost your AI girlfriend.

Can you search “c code to encrypt file”? Also yes.

It’s mostly bias in the training data. Most people aren’t posting mediocre images of themselves online so models rarely see that. Most are also finetuned to specifically avoid outputting that kind of stuff because people don’t want it.

Out of focus is easy for most base models but getting an average looking person is harder.

Unique utility and learning apps built from the ground up vs an api frontend. The choice is pretty obvious.

Today, a simple google search will tell them the correct answer for the quiz question, why bother with discussing the actual answer, the lore and the facts behind it, etc! This is just one example of a creative skill which is rapidly degrading among humans and already taken over by Turing machines.

Or you could acquire more general knowledge than ever before. It’s devalued knowledge economically but it hasn’t stopped anyone from pursuing it.

Machine learning is the same. It has the ability to eliminate many jobs, but that’s not the fault of the tool, it’s the fault of the economic system it exists in.

The paper suggests it was because of cost. The paper mainly focused on open models with public datasets as its basis, then attempted it on gpt3.5. They note that they didn’t generate the full 1B tokens with 3.5 because it would have been too expensive. I assume they didn’t test other proprietary models for the same reason. For Claude’s cheapest model it would be over $5000, and bard api access isn’t widely available yet.

Ten soccer fields you say?

And don’t make a satisfying click

Canada has actually been doing quite a lot of awareness in the past few years. There was the truth and reconciliation commission and there’s a nationally recognized day. Indigenous education has also been integrated into school curriculums in some provinces.

It’s not a ton and can never make up for what happened, but it’s far ahead of Australia who has done nothing from what I can tell.