> Take all studies like this with a grain of salt. A doctor doesn't diagnose a patient by reading a case study. They do it by talking to the patient, examining them, deciding what tests to order, etc. This is a contrived comparison that has little connection to how doctors actually work.

LLM's can do DDX's based on getting the primary complaint and asking follow up questions. It will then provide what tests to order etc. While it can't do a physical exam, they are better than doctors at all other aspects.

It has gone from reasonable amount of ads to utterly insane (especially for iPhone users).

It is now much easier to lose stars and harder and longer to get them back. They want to force users to subscriptions and are being obnoxious about it.

The point of the absurdly large model is to distill the logits to smaller models. Overparameterization makes it much easier to learn the underlying function. Once the underlying function is learned, a drastically smaller model can learn the output distribution (teacher student distillation).

Benefits of homework depend on volume completed (as opposed to assigned) and age (pre middle school have little benefit, middle school benefit from 1 hr, high school 1.5-2.5 hours) and whether the student actually does it (as opposed to copy, etc.)

https://www.readingrockets.org...

As volume increases, students are more likely to skip it or cheat.

Most of the world has widely used public transport that use contactless payment, so it is a fairly obvious transition in most countries. The US lacks decent public transport for most people, and thus they have no experience with contactless payment for public transit, hence the slow adoption.

Since a substantial percentage of the tokens are from Chinese media and sources, it will have the same biases seen in those sources. Just as sources trained on US media will have a US bias.

Of course to the extent that Chinese media publish propaganda the model will learn that propaganda (similar to US media publishing of propaganda), sometimes the US reporting will eventually correct the propaganda though most media don't bother.

As to deliberate censorship - the model itself appears to not be censored but rather when serving the model they have another model that scans the input and output and terminates those responses that are to be censored.

It is a closed source model so there isn't any way to verify their claims.

They aren't claiming a copyright violation, they are claiming a contractual violation - that Deepseek violated the terms of use of their API by allegedly using the API to generate training samples.

People who don't like OpenAI are trying to claim copyright violations that OpenAI 'stole' copyrighted works. Under US law there is 'transformative fair use' - and machine learning models are pretty clearly transformative. So they aren't really the same thing, since in this case copyright isn't being asserted, just a contractual violation.

Now, if a third party had generated the content with the API and published the generated content. Then DeepSeek downloaded and trained on that content, then neither the third party nor DeepSeek would have done copyright violation nor Terms of Use violation.

Just as influenza in the west is usually transfer from birds via poop or saliva to livestock (via the bird dropping food or pooping into feed or water) which infects the livestock, and then the extreme frequency of interactions of ranchers or farmers with the livestock (during feeding, slaughter, etc.) results in the transfer to humans.

LIkewise a bat likely pooped or dropped food into the feed troughs or water troughs of an 'exotic animal' farm and the similar high rate of interaction with farmers/ranchers resulted in COVID transfer.

It is those extremely high rates of human animal interactions that provide an opportunity for exposure to enough mutation variants that a tranfer to humans can occur.

The number of interactions of a worker with an infected sample in a lab in such a way that they could inhale or transfer hand to mouth are so extremely low that the 'escape from research facility' theory is rather absurd.