Hacker News Reader

Comment by sinuhe69 | Hacker News Reader

What an irony! The Bostrom Institute has to close in the midst of the spectacular rise of AI, and the all-important questions of AI risks and human value alignment are more hotly debated than ever.

I hope that Bostrom and his colleagues will continue and deepen their research, because humanity needs such insights now more than ever.

View on HN

How does one research and generate insights for what does not exist yet? Is there a framework? At SXSW this year, there were a couple of talks about forecasting, as a tool for futurism, but where ethics are concerned, I’ve not seen a good template. Maybe all if this is obvious, yet I’m curious so I’m asking.

I'm not expert and struggled to read Bostrom's Super Intelligence, but my interpretation was it started with a lot of broad, hand-wavy factors & historical interpetations, layered on some pretty tenious projections as inevitable and wanked off with deep thought experiments. I get this sort of open-ended exploration has value, but I'm not sold on it's "prioritized value" when compared with other areas and directions.

I just finished reading it.

It's extremely repetitive; he could have written a book one-third of its length without leaving anything out. There's some good stuff in there, but I was sort of annoyed that the author forces you to read everything three times. Kinda disrespectful to his readers' available free time.

It was literally under the department of intellectual wankery. Them shutting it down is the pot calling the kettle….

> How does one research and generate insights for what does not exist yet?

It's Zeno's paradox of research: we cannot think about what does not already exist, therefore nothing new can ever be brought into existence!

I call this "nothing can ever happen for the first time". One often hears it in passive-aggressive arguments against renewable energy.

it seems plausible that all new things are brought into existence unintentionally. that is, you tried to create something you imagined, but what you actually created was something else. certainly there are many examples of thinking about renewable energy that were completely wrong because they were based on presumptions that no longer hold, or in some cases were already wrong but in a nonobvious way

as my wife points out, we can't even imagine any of the things that do actually exist, only drastic simplifications thereof

Never heard that in that context. Can you cite an example?

I constantly see the argument "no country has ever been powered by solar and wind", implying that it therefore can't be done, and that one should use nuclear instead (never mind the same is true of nuclear; not even France is fully nuclear powered).

> never mind the same is true of nuclear; not even France is fully nuclear powered

This seems somewhat bad-faith: nuclear does supply the majority of France's power, and since nuclear is a "base load" part of the mix, it would be inefficient to get 100% of power from nuclear rather than a peaking-friendly mix that includes e.g. hydro and gas.

Focusing only on the grid is bad faith; if one includes non-grid energy use, France doesn't even get 50% of its energy from nuclear.

What's the non-grid energy use?

Transportation, industry, maybe heating.

Are you confusing electricity generation with electricity use? If we're counting off-grid electricity generation, then we're talking about A) critical infrastructure like utilities or B) off-grid homes or communities which typically use solar panels, small wind turbines or gas/biogas to generate their own power. The off-grid portion most likely make up a tiny portion of France's energy mix, which is dominated by nuclear (>50%).

No, I'm talking about energy use that doesn't involve electricity production at all. It potentially could be electrified in many cases, but that doesn't reduce the hypocrisy of the selective "if it hasn't been done it can't be done" mindset.

How is that relevant in this context, though? Why is it bad faith when we're talking about grid-supplied electricity generation? We need electricity generation, and most of France's is nuclear. It seems uncontroversial.

It's because the "if it hasn't been done, it can't be done" argument applies to nuclear displacing fossil fuels. Remember, the decarbonization goal has to be all uses of fossil fuels, not just their use for the grid.

I honestly think we have crossed wires. I don't think anyone here is saying things can't be done based on history.

I have repeatedly and literally seen this argument applied to renewables. I've even seen it on HN.

Is FHI even needed anymore? His ideas clearly won and took hold of a large group of people. AI doomerism is the default stance most people have on AI. e/acc makes up a tiny fraction of popular discourse today.

Geez if this is what it all looks like for AI to be unpopular I can't even imagine what its going to look like when everyone is actually into AI.

Reply on news.ycombinator.com