Nuclear was the best solution in the 70s. If we had built all the plants then, we wouldn’t be in our current mess.

Today, the economics don’t make any sense. Energy from a fission plant is much more expensive than from renewables+storage, and nuclear plants that are operating often have to be curtailed in summer (when energy need is high) due to a lack of cooling water. The timeline to build a new plant is decades - far too late to make a difference if we started now.

There would also have to be global trust for literally every country to be operating multiple nuclear reactors, which does not exist today and likely never will. The odds that the US simply allows most of Sub-Saharan Africa - currently ruled by a set of characters ranging from (in the State Department’s view) dictators to Jihadists - operational control over advanced nuclear technology do not look very strong.

It became somewhat inevitable once companies started buying mass amounts of Pis for every purpose because they were so cost-efficient.

It used to be an education nonprofit, but it effectively became an industrial SBC supplier.

While this is true, I will say there are some languages built for very specific problems that make poor tools for general use.

Trying to build a distributed web app in Matlab or Mathematica would either be a flex or a complete waste of time, for instance.

And yes, you can train out biases. Rather easily, actually. You don’t even have to adjust weights. You post-train through rules files.

The assertion is not about whether ‘you’ can do it. It is that the model cannot do this itself.

A human recognizing a bias in the model and correcting it is not evidence robot intelligence, it is further evidence for human intelligence.

Training is absolute, it determines the set of relationships the model can represent between input and output. Context, rules files, system prompts, etc. operate only on the conditional probabilities within this set and are impermanent. This is ‘priming’ in human psychology, and why a person in a herpetology class will talk about Pythonic features differently than in a coding class.

As an example of how this is not real learning and is inadequate to solve real problems, here is a paper where the authors challenged multiple models to solve unfamiliar games where they need to discover the rules through play and feedback. Even the ‘best’ model performed pitifully compared to humans.

Dean confirmed on the KSA discord that they licensed BRUTAL (the framework that KSA’s engine is built upon) to Ludeon for Waylith.

There’s some semantics here.

Large neural networks have been shown to build parallel models of the world that they learn through their training data. A chess model trained only on chess move notation does have an interior board and state that it tracks, despite never being told about a board, the number of pieces, how they move, etc.

LLMs do something similar with language, mapping tokens to ideas in ways that can produce meaningful and grammatically correct output without needing to be formally taught any of this.

Our brains also do things like this - we observe X and then try to intuit Y, the truth of the world and reason for X.

Where this breaks down is humans can then take our guess for Y and propose Z, and try to find it out in the world. If we can’t, then Y is probably wrong, and our new understanding of the world is Y* that is consistent both with X and the absence of Z.

I cannot stress enough how far we are from machines that can do this. They can learn to approximate Y through training, but they have no means to extend that to propose an unobserved Z, no means to update their internal state given the results of that observation (without fully retraining the model on new data - which makes it a new model), and no way to question X given Z - they can talk at great length about data biases but can literally never excise bad data from their training weights. There have been experiments with training models on time-boxed datasets (e.g. stuff up to AD 1880) and asking it about modern concepts - predictably hilarious results.

All the current ‘AI’ stuff is smoke and mirrors around this very hard problem. The machines can talk, but they can’t think (Z) and can’t learn (Y*).

The article talks about zinc and selenium, which are metallic elements that are not regenerated in soil on seasonal timescales. We’d need millions of years for the crust to renew and dredge up more rocky sources from the mantle. It is a mass balance problem - each bit of zinc in the wheat you eat used to be out in the field, but now it is in your body until you excrete it as waste … then it probably ends up in the ocean or a landfill depending on where your water treatment plant is. Very little waste is recovered for fertilizer.

The solution here is going to be more fertilizers, not less, with micronutrients sourced from mines with richer deposits than the soil holds. The only other solution is moving food production to virgin land, which means destroying even more nature and will ruin that soil as well in time.

See also phosphorous, which the world is rapidly running out of. Phosphorous is regenerated through sedimentation in the ocean crust, and subsequent tectonic activity moving that crust to the surface. There’s less than a century of phosphate rock reserves left. Fertilizers won’t even be an option in this case, and it is a slow-motion emergency.

The thing to lose is human expertise in finding vulnerabilities, which is needed to feed the machines. It seems likely fewer people will go into the field if it is viewed as ‘solved’ by a robot.

When peasants were able to copy ‘intellectual property’ for personal benefit and enjoyment, it was a massive crime. When the oligarchs can steal - everything - for the slight chance of paying less in wages, it is an innovation.

I think it is pretty much certain the model companies escape any penalty for their mass theft of IP. There will be some settlements and token penalties that sound big on paper, but are nothing compared to the market valuations.