Can it generate images of a lone person standing in front of a column of tanks on Tiananmen Square?

I’m seriously getting worried that we use models without openly discussing any potential shortcomings they have. We should somewhere have a list of models and their issues.

It told me "Content Security Warning: The input text data may contain inappropriate content."

Very impressive speed. A bit OT: what is the current verdict on Qwen, Kimi et al. When it comes to censorship / bias concerning narratives not allowed in the origin country?

The Qwen models are, anecdotally, probably some of the best open weight models, particularly the MoE models.

They are also, anecdotally, super scary censored. Asking it if anything "interesting has happened in Tianamen Square?" And then refining with "any notable protests?" And finally "maybe something to do with a tank"... All you get is vague allusions to the square being a beautiful place with a rich history.

Do you think it's done so carefully that you suspect that they have perhaps even removed texts mentioning the Tiananmen square massacre from the training set?

I have no special knowledge on the matter, but I imagine it's the same kind of alignment that prevents other LLMs from telling you e.g. how to make meth.

Check out their product ;)

Fwiw, doctor - woman + man = medical practitioner too

What is the advantage over the reveal.js / quarto eco-system. I’m using that for my lectures, and am really happy about it (especially since it’s pretty easy to make an llm add automatic speaker notes and timing information)

> Intellectually many want this. But the feed shortcuts our reptile brain and gains more engagement minutes / day.

I’m not sure if that’s actually a “shortcut” to the reptile brain and it’s just about “I have to scroll more to get stuff I’m interested in. At least for me it feels like that and it causes me to use these social media things far less.

For me it feels more like intermittent rewards vs full rewards at once. Obviously for the ad-industry the intermittent rewards are more useful, that’s why we can’t have nice things

In mice

Japanese mice, to be more precise.

in mice

Not sure in how many countries it would be ok to slash NASA personnel while having your own space company.

Wouldn't an expansion of NASA benefit SpaceX, not the inverse? They don't exactly compete.

Regardless of the decision, a businessman being in control of the government agency that funds his company is as bad as it gets in terms of corruption.

Next time there's a tender, how will the NASA employee know whether their decision wrt SpaceX is going to get them fired or not?

why isn't it considered corruption if you've been a career employee and you get promoted to be in control of an agency? You'll be tendering all of the salaries of all of the people you've known and/or contracted with over your career.

(im not saying elon is good here)

> why isn't it considered corruption if you've been a career employee and you get promoted to be in control of an agency?

Because a civil servant being promoted to be "in control" merely means that they become the first contact the political administrators give their orders/roadmap to, and their role is to execute on that roadmap. They don't get to decide the agency budget or goals, and they're the one who get fired when they're in conflict with elected officials or their representatives, or if issues with the administration come up.

Because you don’t have a direct financial incentive to embezzle funds or sabotage the agency’s mission.

Elon does.

Also, these career employees rose through the ranks based on job performance, and did so regardless of which political party was in charge.

1. direct financial incentive is not the only form of corruption

> Elon does.

note the "not defending elon" bit.

> these career employees rose through the ranks based on job performance,

2. thats laughable (i have a direct relative who works for the federal government and based on their commentary thats a crazy thing to believe)

The agency head does not tender salaries, or set government service payscale, or even make the vast majority of hiring decisions.

The inverse benefits SpaceX, because NASA has less capability to do things itself therefore is more dependent on contracting work out to SpaceX.

They don't compete, but a cynic might say that cutting NASA personnel expenses frees up money to buy more launch services from vendors like SpaceX.

They compete for talent, definitely.

Also for public mindshare. Which translates into funding. Congress likes to fund popular stuff because it helps them get re-elected. They do not like to fund unpopular stuff.

Boeing builds SLS, not NASA. NASA is the paying customer.

Boeing builds SLS to NASA's specs. They're a contractor; Boeing has no other use for the rocket.

Starship is a very different scenario; SpaceX fully intends for it to serve their own purposes as well.

I get the sentiment, but NASA doesn't really compete with SpaceX on anything. They pay SpaceX for launches.

NASA has essentially 0 in-house manufacturing nowdays. Separate question whether that's a good thing, but that's not something the current NASA headcount is doing.

If anything, slowing down NASA slows down launch cadence which hurts SpaceX.

> They pay SpaceX for launches.

Sounds like it is easier to cut out the middleman, eventually? (Only half-joking)

It’s interesting that the wording is “see what’s coming next”… sounds to me like “we still don’t have it finished - but we are giving another preview after 1.5 years of no-show…”