First of all, props to student 22 who clearly didn't cheat on the midterm and managed to do better on the final, but it looks like about 85 out of 89 students cheated.

With cheating being so easy, professors really need to be having in-person tests to see if the students are learning anything. I wonder if giving students the exact same test out of class and then in class would show who cheated more clearly

Anyone that knows US history knows the parties have essentially swapped so the people who would have been a racist southern democrat 100 years ago would now be racist southern republicans. Ignoring this and continuing to argue that the modern parties share the values of the past is at best ignorant but most likely a bad faith argument made by someone embarrassed by the position of their fellow conservatives.

Then you'll be much more at home with the left. Despite some extreme views, that's a view much closer to mainstream Democrats than Republicans. On the conservative side you have people that are openly racist (Trump for one was found by a court to be racist, white supremacists are a part of the right wing) and sexist.

There are some people on the left looking at also correcting injustice of the past (versus the right wing, looking to return to the injustice of the past).

Conservatives: People being treated humanely and equally? That's too radical a change!

What a bunch of clowns

That's an idealist but somewhat pointless list of requirements to keep you from ever trying a tool.

That's pretty much based on the training set and the prompt. You can prompt it to behave how you like

Not really possible to know where a model was trained so an easy excuse to ignore it.

Another impossible requirement. All models will include material from public domain or licensing deals that won't explicitly include the people who made the material

At least this one is somewhat possible. Obviously it excludes all commercial models because it requires you to know the future, but at least local models fit the requirement.

Florida already has major problems with hurricanes and sea levels rising is devastating for a state that's mostly flat with swamps.

They can stick their heads in the sand about it or they can work to slow the problem, but the anti-science government doesn't seem to care.

Don't forget to add "Make no mistakes" so you know it's safe

I don't think the engineers they're hiring are working on those systems either

That sounds very dependent on the car shape, driver size, seat position, and road. Might be good for some general guidance but no serious driving instructor would suggest that.

Can't forget the bots!

Pretty easy to put lower limits on that sort of thing or even exclude a primary residence

Microsoft is perfectly capable of introducing more flaws without the help of AI

Don't think of "Fable" as an LLM, think of it as a system which utilizes the LLM "Mythos". Alone, the technology doesn't have that capability, but Anthropic isn't giving people direct access to Mythos, they're running communications through a filter which is probably another LLM trained to detect exploits

Seems like they've put very conservative limits on it which err on blocking usage. According to the blog post one group managed to jail break it over a long session but no one else has been successful

I'm guessing Musk had SpaceX acquire xAI and then the people that actually know about business at SpaceX said "This Grok product is a big money pit, but there's a shortage of hardware that Google and Anthropic will pay loads to use"