- Why a tokamak? Two reasons. One is that the UK has been on the cutting edge of tokamak research for years, so itâ(TM)s what they know and can contribute to most. Two, because itâ(TM)s the safe bet. High temperature superconductor based reactors are at a point where itâ(TM)s all but certain that theyâ(TM)ll produce power in the next few years. Other approaches like helionâ(TM)s are far less certain, and a whole new research avenue.
- why a supercomputer? Because while plasma control is incredibly difficult, it is proving to be tractable. The UK and china have both been making strides in understanding how to do it, and have managed to run their reactors for longer and longer. What it does need though is powerful computers both to research how to do it, and to actually do the modelling to control it.

Because ITER is a boondoggle thatâ(TM)s outdated before itâ(TM)s even finished. Small high temperature superconductor reactors like the ones the UK are planning are far more feasible to build, smaller, cheaper, easier to iterate on, and much further inside the area of the graph where fusion works.

No, what was needed was the discovery of rare earth, barium copper oxide superconductors, and the necessary research to figure out how to build large magnets with them. Thatâ(TM)s why commonwealth fusion systems in the US is also predicting the same timeline.

For what itâ(TM)s worth, the UK has pretty much always been on or near the cutting edge of fusion research. Itâ(TM)s in no way surprising that theyâ(TM)re one of the first to be saying âoeyeh, we think we can actually build a working reactor nowâ.

Yeh, and honestly, if they lose this, then America is screwed. You have exactly 0 constitutional rights if this doesnâ(TM)t succeed.

No, now can the Supreme Court reverse this ridiculous decision. *this* is the kind of dumb shit that genuinely does get your famed rights and freedoms taken away.

There are however a *lot* of texts in the training data saying "launching nukes would be a terrible, world ending idea". I find it rather puzzling that the AIs don't attempt to avoid using them at all costs. It's a shame we're so bad at analysing why LLMs make the decisions they do

I would imagine that the web site owners would be told âoeyouâ(TM)re not allowed to send any naughty data if the OS doesnâ(TM)t have support for the APIâ.

Honestly⦠this sounds way better than the current state of affairs where you have to share your government if or credit card with loads of random 3rd parties with what I bet is all the security of a wet paper bag.

Given how long itâ(TM)s taken to get to large, online models that are capable of decent quality coding, I suspect itâ(TM)ll be a while before we have local ones that can do it. Either we need a significant breakthrough in model efficiency, or we need much much faster hardware with much more ram (hahahaha)

Yes and no - they claim to predict things *over* the next 50 years. We can check them each and every second until 50 years passes with increasing confidence (or lack there of) in them. The predictions made in the 80s and 90s have so far turned out to be largely accurate, so seems like weâ(TM)re more in the science box than the alchemy one.

Really? Which climate predictions are we talking about here?

Hansenâ(TM)s 1988 models inaccurately predicted emissions levels, but when adjusted for actual emissions levels using the same methodology turns out to be fairly accurate to reality.

The IPCC report from 1990 predicted 0.3Â of warming per year, which tracks well with the 0.2-0.3Â per year weâ(TM)ve seen.

Early predictions of arctic ice melt predicted that the
Volume of arctic sea ice would have fallen by 35% by now, it has fallen by 40% - pretty accurate.

1990s predictions of sea level rise predicted 18cm by now, its risen by 20-25cm - so somewhat conservative but the right trend.

The one prediction that I think it would be possible to point to as âoewrongâ is the idea that freak weather events would increase as sea temperatures rose. The rate of such events has turned out not to rise, however, the severity has risen instead, so this one is a bit off but not significantly.

Another way of phrasing this:

Two projects aim to come up with new models. One is using vast amounts of data and back propagation to learn the values of various coefficients in an existing model. Another is using vast amounts of data and back propogation to come up with a new model that is a vast array of linear equations combined with sigmoid functions.

Phrased that way, it becomes clear that the former is likely to be far more useful for extracting understanding of whatâ(TM)s going on, while the latter is likely to be much more accurate, but impenetrable in terms of understanding.

These are people who were facing serious prison sentences - like, decades. They knew they were innocent, but had had everything they owned taken from them, their reputation dragged through the mud, and were going to prison, and were completely powerless to stop it. I am in no way surprised some of them committed suicide.

Pretty sure weâ(TM)ll get at least a working prototype fusion reactor by 2030. SPARC is well under construction, and thereâ(TM)s a few other promising alternatives out there.

1. Thatâ(TM)s great and all, but can we identify which rocks wonâ(TM)t move for the next 100,000 years? We can certainly make educated guesses, but our understanding of how the planetâ(TM)s surface moves is still pretty immature.

2. Great, you protected it against geology, what about protecting the people who are mining in 5000 years, and have no idea what they just hit? Thereâ(TM)s no guarantee at all that theyâ(TM)ll understand any of the warnings we try to put in place, or know what radiation even is.

3. Great, but can you commit to guarding it from anyone who wants to make a dirty bomb for the next 100,000 years? I doubt it.

Just finding some stable rock and burying it doesnâ(TM)t magically solve the problem.