Here's how I look at it: generative AI is dsigned to create plausible-looking output in response to a prompt. That's how the lawyer who submitted a ChagGPT-generated brief got caught. The references the AI generated for the brief looked plausible enough to pass a cursory inspection, even by an expert, but if you looked them up they didn't exist.

This wasn't a failure of the AI; the AI did exactly what it was designed to do. It was the fault of people who relied on it to do something it wasn't designed to do. Surely someday soon a *lot* of the work of peer review will be automated by AI before a paper actually gets reviewed by a human, to avoid wasting reviewer time with obvious shortcomings.

Systematic reviews, including meta-analyses, are very common. Since roughly the 1980s systematic reviews have been considered the highest possible tier of scientific evidence.

This is because of certain facts about science that outsiders often find shocking: (1) complex questions nearly always have contradictory evidence and papers taking opposing views of issues, particularly early on; (2) every paper, no matter how good, has methodological shortcomings if not outright errors; and (3) many novel findings never get replicated. This means individual papers are almost useless for proving anything, at least without putting them in context.

That's what systematic reviews and meta-analyses do: they put individual studies in context of what other researchers are finding. Unlike some internet rando citing papers to prove his pet theory, a reviewer can't cherry pick papers based on what they find. There are rules that ensure papers can only be excluded for objective reasons that apply to all the papers on the topic.

In this particular paper, the authors did not find evidence that deforestation and forestation were drivers of disease. An activist writing a polemic wouldn't come to that conclusion.

The brain starts with the semantics. Some are innate, others are learned, but the semantics is always first. The syntax is then layered on top of this. This is why the high-intelligence end of the autistic spectrum is linked to delayed speech followed by a very rapid process to complex speech. The semantics is being built to a far higher degree, the syntax is postponed until the last possible moment.

AI, as it currently exists, needs a very very large number of examples, far more than the brain by tens of orders of magnitude, and hallucinates far more, because ALL it knows is the syntax. There is no handling of the semantics at all.

This approach can NEVER lead to actual intelligence of any sort, let alone superintelligence. They are solving the wrong problem. And that is why they fail, and why they will only ever fail.

If you want actual intelligence, the syntax must come LAST. And the modern breed of AI researcher is simply far too stubborn and arrogant to fathom that.

I'm sure she'll be happy to strap you in front of a camera.

Next time I'm reincarnated I'm going to opt for being a member of the privileged class.

The company management is pointing the finger at workers, and they're right to, just as long as they point the finger at themselves too.

These kinds of problems start at the top. If management demands workers do the impossible (or at least the wildly implausible), they know that reports of success are going to be fraudulent. The question is, are they goign to get away with it?

I really think the main argument *for* carbon offsets is that it *potentially* can harness free market mechamism to *efficiently* reduce emissions. This would be in contrast to a pure government mandate that everyone cut their emissions by some percent. The problem is that the marginal costs for industry X might be prohibitive; on the other hand industry Y could easily cut more. So why not have X pay Y to cut more than required? This *internalizes* the external benefits of extra reductions for Y.

Of course, it's very easy to screw this up, starting with letting people get away with fraud. But if you allow fraud in *any* market, that undermines the efficiency of the market. If you are going to get the entire economy to reduce emissions by some set goal, you need some mechanism to distribute those reductions so they're made where it's most efficient, and financial efficiency is one thing the free market excels at.

So let me get this straight. If there were *real* carbon offset credits, that would in your view be *just as dishonest* as fraudulent ones?

how?

If they want us to try their coin, they should hand out some free samples.

Uber drops drunk driving rates.

Nanny state garbage is not the answer.

Italy and Japan have shrinking populations. We would too, if it weren't for immigration. However our population growth rate is still low, and if it were any lower we'd be facing serious economic and social challenges. Sure, a shrinking population would drop housing prices, but we are far from having so many people there isn't space to fit them. Our real problem is seventy years of public policy aimed at the elimination of "slums" and the prevention of their reemergence.

If you think about it, "slum" is just a derogatory word for a neighborhood with a high concentration of very affordable housing. Basically policy has by design eliminated the most affordable tier of housing, which eliminates downward price pressure on higher tiers of housing. Today in my city a median studio apartment cost $2800; by the old 1/5 of income rule that means you'd need an income of $168k. Of course the rule now is 30% of income, so to afford a studio apartment you need "only" 112k of income. So essentially there is no affordable housing at all in the city, even for young middle class workers. There is, however a glut of *luxury* housing.

In a way, this is what we set out to accomplish: a city where the only concentrations of people allowed are wealthy people. We didn't really think it through; we acted as if poor to middle income people would just disappear. In reality two things happened. First they got pushed further and further into the suburbs, sparking backlash by residents concerned with property values. And a lot of people, even middle-class young people, end up in illegal off-the-book apartments in spaces like old warehouses and industrial spaces.

Since communism is, by definition, classless, you cannot both be classist and communist.

Racism is, again by definition, incompatible with abstraction.

My conclusion is that you're pretty shaky on definitions.

Currently rated "overwhelmingly negative" on Steam. I've never seen that before.