Well you still won't know until you have an actual product that's viable to the consumer. Until then it's just a guess. You can't make reasonable estimates about something you don't know right? Technological development can be forecast, even when we don't know it. And the forecast for net energy from fusion is short.
And has been short since the 50's, hence the joke.
I doubt we're going to come to an agreement on this. I've worked in product development long enough that I know that you can have reasonable estimates on how long a prototype will take to bring to market, assuming that you have a good understanding of your technology. You can be wrong, but these things can be planned out with reasonable allowances for the unforeseen or unexpected and you can explain how and why you chose those numbers. Once you have a functional prototype, the path to final product is usually more about how much time and money are you willing to invest in getting there, not is it possible at all.
I had a product proposal once where I knew from practical experience that what the customer wanted was possible, it needed some work on tweaking the design to their specific application and some testing to verify that it would function as required. Estimated cost to do the work ~$40,000, based on the materials I would need, the time I would need to spend on it, a reasonable appreciation of how much extra time might be required should certain things not work out as I expected and some profit for the company. As part of my proposal, I had to provide reasoning and justification for all of those.
(The customer wanted it done for under $3,000. Thank you very much, but we will not be proceeding with this project.)
Try doing the same for "fusion in 30 years". The problem is that you have a process that works on paper, but has never been observed working in practice. You're not even at the prototype stage yet, so you don't even know whether this particular technology is on the path to a final product. Until you have a working prototype, estimates are based on assumptions that your experiments will function more or less as expected. Experiments that you're doing precisely because you're not sure whether they will function as expected. Estimates of what might be required if the experiments fail is even harder, because you sort of need that data in order to map out the development space to see what might be a viable solution. You don't have the luxury of "I want to try this cheap and easy thing to see if it works, but if it doesn't I know I can fall back on this more expensive but proven solution". It's possible that no amount of time and money will make this particular design functional, hence why any planning is more optimism than actual data.
You can plan this stuff out if you want, but the dependencies and the variability means that you end up with something that is more like "fusion probably in 30 years, but almost certainly somewhere between 25 and 150 years". The spread on the reasonable estimates gets so large that I question the meaningfulness of giving a number at all, as it tends to just be misleading. Whereas with a traditional project where a functional prototype exists, it's more like "commercially viable product in 3 years, or 4 if everything goes wrong". The absolute worst case scenario is something like the F35 where it ends up being really late and really over-budget, but there was still never any real doubt that it
could get built because there was a functional prototype before they started pinning time-lines on it.
Yes, for this specific application. But this specific application is also benefiting from AI technique refinement.
"Yes" is not an answer to "when you say AI, what do you mean for this specific application?"
I'm asking what you mean by AI so that we can have a reasonable discussion without me putting words in your mouth. If you don't want to talk about it, just say so.
