h/t The Register; President Biden recently urged the DOE to produce viable commercial nuclear fusion by 2032. So the DOE is offering money to AI researchers, in the hope that the magic of AI can help make the magic of nuclear fusion a reality.
…
SUMMARY
The DOE SC program in Fusion Energy Sciences (FES) hereby announces its interest in applications in the areas of Machine Learning (ML), Artificial Intelligence (AI), and Data Resources for fusion energy and plasma sciences. The goal of this FOA is to support multi- disciplinary teams aiming to apply advanced and autonomous algorithms to address high-priority research opportunities across the FES program. Applicants are encouraged to propose research in new systems for managing, formatting, curating, and accessing experimental and simulation data, provided in publicly available databases. Of high programmatic importance are approaches that support the realization of a fusion pilot plant on a decadal timescale.
SUPPLEMENTARY INFORMATION
Program Objective
The FES mission is to expand the fundamental understanding of matter at very high temperatures and densities and to build the scientific foundations needed to develop a fusion energy source. This is accomplished through the study of plasma, the fourth state of matter, and how it interacts with its surroundings. A 2018 report by the Fusion Energy Sciences Advisory Committee (FESAC) on “Transformative Enabling Capabilities (TEC) for Efficient Advance Toward Fusion Energy” [1] included the areas of machine learning, artificial intelligence, and mathematical control, as part of its Tier 1 “Advanced Algorithms” TEC recommendation. Within fusion energy research, AI/ML impacts key research problems from applications requiring rapid
processing of fusion data for active plasma control to exascale applications that use high performance computing resources to simulate multi-scale processes impacting fusion plasmas.
Recognizing the potential of AI/ML and data science, FES and ASCR co-sponsored a community workshop in 2019 to identify areas in fusion science where application of AI/ML can have transformative impacts, and to identify unique needs, research opportunities, and associated gaps that can be addressed through targeted partnerships. The final report [2] identifies several high-level priority research opportunities that span DOE’s Congressionally authorized mission- space in fusion energy and plasma sciences. In 2020, FES supported a broad range of research activities in the areas identified in the report. This FOA continues to encourage applications in previously supported areas while also expanding the range of supported activities.
Recently, DOE initiated an agency-wide activity to accelerate fusion energy research, development, and demonstration activities to realize a commercially relevant fusion pilot plant (FPP) on a decadal timescale, an effort known as the Bold Decadal Vision (BDV). This followed the 2021 National Academies report “Bringing Fusion to the U.S. Grid,” which describes innovations and requirements needed for a FPP [3]. A key component of the BDV is a new milestone-based fusion development program, authorized in the Energy Act of 2020, that aims to accelerate fusion energy RD&D in partnership with the private sector [4]. The milestone-based program includes a range of FPP design activities. Efforts supported under this FOA can play a supporting role in these activities and other SC/FES programs by providing broad access to curated experimental data and theory/modeling-based analysis.
…
Read more: https://science.osti.gov/fes/-/media/grants/pdf/foas/2023/SC_FOA_0002905.pdf
The DOE has been doing a lot of work with Google, training their deep mind AI to learn what interventions are required to keep the fusion plasma stable in tokamak reactors.
AIs are good at solving some kinds of optimisation problems, particularly when a good model is available for the AI to use to evaluate its own skill.
Optimising the laser configuration for say an inertial confinement fusion reactor, or the magnet arrangement for a tokamak or stellarator, are problems which can potentially be tackled by AIs.
But the DOE has tried this, and still haven’t cracked the problem. So they want to talk to someone who can give them new ideas, someone who can provide them with hope of developing an AI with capabilities beyond everything they’ve tried to date.
Given the pressure from President Biden to produce results, I suspect there will be a lot more money on offer than $33 million, if anyone comes up with a credible idea they haven’t heard or tried before.
Update (EW): My favourite AI demo – Travelling Salesman. This example demonstrates how AI can be used to find an optimum solution, in this case the shortest line which joins all the dots.
Still waiting for cellulosic ethanol.
I prefer grain ethanol, however, it’s hard to beat a good Cognac, generally the more expensive the better.
I prefer bourbon- but YMMV.
Still waiting for AI.
That’s the only hope, as “I” without “A” disappeared from a woke government.
Al has spoken.
Didn’t you see “An Inconvenient Truth”?
The Goracle is never wrong!
The DOE would have better luck funding development of a perpetual motion machine.
They have. See the $25 million they put into Skyonics carbon capture.
It would be interesting to see the AI software iterate to the point where it shows that the best confinement would be in the vacuum of space and confined by its own gravity – i.e. another sun!
Self-confiement machines (without the vast mass of the sun) ave been built and achieved twp of three milestones to fusion. Example: LPPFusion and its Dense Plasma Focus machine
Telling Greenpeace and their ilk to sit down and shut up over fission would yield power on the grid much sooner. Most of their concerns were over preserving the Soviet Union, rather than any real environmental effects.
Yeah, that is quite cynical, but given the level of purely bogus claims the anti-nuclear movement has made over the past 77 years, other issues clearly dominated their motivation.
AI is at heart a big pattern recognitIon machine. But that does not even in concept apply to physics problems like random tokamak plasma pinch caused by chaotic magnetic boundary instability— for which there is by definition no pattern.
The assertion that a hammer applies to a screw does not inspire confidence in the energy ‘science’ of the US Gov.
Yeah, I’d rather see NASA put a certain president on the moon by the end of 2023.
Why wait?
Good point.
Because an SLS launch costs north of $US2B (yes, B). Bascially only one launch per year.
But we know more about the Moon than we do about the Oceans.
One anchor would be a lot cheaper and could handle more than just one.
All that’s needed is for Biden to issue an Executive Order for DOE to build a grid scale fusion reactor by 2035 and then expand the application to the entire grid by 2050. If that’s not enough, I’m sure congress can pass a bill to mandate that it be done. After all, it’s what they’re doing with wind, solar and batteries. We all know any problem can be solved with enough other peoples money – according to liberal/socialist doctrine.
Yes, AI is just another computer model/simulation that will output what it’s programmed to output.
Oh brother.
I stole this without shame from the “signature” of a Microsoft expert on some help center message board.
“AI: Artificial Intelligence or Automated Idiocy?”
I think that poses the right question, and that the answer is a lot closer to the latter than the former.
Why automate something they have perfected? (idiocy)
I already have a design for a practical fusion reactor, but space limitations prevent me from providing it here. Sorry, DOE.
Sounds good to me but, who are we going to get to understand the AI.
Please God, not the ignorant and destructive tools we currently have feeding at the govt trough in the name of science.
All my software developer friends loved the Terminator movies, because the story of how Skynet turned hostile was consistent with how real computer bugs happen. Just saying.
But actual AI, machines that reason and are self aware, doesn’t exist. What we call AI today is just a bunch of sophisticated algorithms. They just perform their programmed functions, and can’t “think” outside those functions.
Yes and no. Some work is being done on giving AIs the ability to extend their own capabilities, such as NEAT, where the neural network modifies its own topology to give itself new capabilities. The NEAT creator has also done some groundbreaking work on goal seeking, which might eventually lead to strong AI.
Chess engine AlphaZero played in ways no human player ever thought of.
How do we know that?
But did it decide to stop playing chess and do something completely unrelated, without being told to?
So, not much different frin the average Liberal, then?
I’m not afraid of supercomputers. I’m afraid of their human interpreters.
FWIW Skynet wasn’t a bug. Reaching self awareness frightened humans attempted to deactivate it… bad move. So how about we don’t do that this time. O.k… 🤓
Hehe – the bug was how the computer reacted to authorised human software technicians reaching for the off switch after unusual activity was detected ;-).
We have had at least a half century of warnings of AI from scifi movies (and books) like Terminator, 2001 a Space Odyssey, The Forbin Project, I Robot, etc.
We really need to stop doing this.
It’s a feature.
Let’s not forget VIKI and HAL!
Or WOPPER!
WOPR
They will just pretend that they have made the breakthrough and sceptics will be silenced.
Coal plants will be renamed “nuclear fusion plants”? I love it 🙂
They just did that about a week ago didn’t they?
“if anyone comes up with a credible idea they haven’t heard or tried before.”
i hear Sam Bankman Fried has a lot of good ideas. And a lot of time on his hands, too.
Clearly the DOE should be using blockchain to advance their fusion research /sarc.
Let’s put it in perspective: How many windmills does $33M buy?
I was an CompSci undergraduate in 1980, We had a lecture in AI and most of the attendees didn’t realise until 15 minutes in that she wasn’t talking about Artificial Insemination
“Hal, open the air lock”.
‘Sorry Dave I can’t do that’
“Is here a reason you can’t do that Hal?”
‘Yes Dave, I think, therefor I can’
Great comment on a classic early case. Minor purist nerd snipe…Dave actually said “open the pod bay doors, Hal” and the first time he actually said please. Another similar vintage and fine tale was Colossus: the Forbin Project. Perhaps intelligence isn’t the only attribute we need in a sentient entity.
The only pressure from Dementia Joe is to produce an ice cream cone or at least a pudding pop.
As for his puppetmasters, there’s no pressure to produce an actual breakthrough either. All that matters is that there should be a steady stream of fake propaganda “breakthroughs” to string the faithful along.
Magnetohydrodynamics pretty much says that these systems cannot work, due to magnetic chaos interactions. So the “science”says there is a problem, but idiot “believers” say that computers can fix it, with no understanding of the problem? Collecting the data on the required scale should be interesting, particularly as no experimental data is properly available (a tiny time slice is useless). Give me some cash too please!
Project Pacer figured out a way to make inertial confinement work – detonate a H-Bomb in a salt cavern, then suck heat out of the salt for the next few years. Rinse and repeat. It is not a sensible way to produce electricity, but as a proof of concept it demonstrates there probably is a path to make inertial confinement work, given that one feasible concept has already been identified.
What does “make inertial confinement work” mean to you Eric?
There are those who believe that fusion can be said to “work” if for any period of time, more energy comes out in the form of waste heat than is put in in the form of useful electricity. But of course they’re willing to cheat and ignore the inefficiencies of the whole system and just talk about the energy that theoretically impinged on the fuel pellet.
The same people don’t see any issue with saying that such a system would “work” if it costs $10 billion to build, lasts for six months while producing 8TW-hrs of electricity and several tons of highly radioactive waste. In other words, they imagine that producing electricity at a cost of $1.25/KW-h “works”.
When I say it will never work, my definition is that it will never be commercially viable.
Pacer is interesting because it is close enough to viable one could imagine extreme circumstances in which it might be used – say a quick and dirty power system for an off planet colony, or emergency power after a major catastrophe, like a big comet strike.
But yes, nobody is going to apply for a civil license to detonate nukes.
You missed or avoided my point Eric.
Manmade fusion has worked since the early 1950s in the form of hydrogen bombs. If the question were whether you could design a kind of enhanced geothermal power plant by storing heat in molten salt generated by H bomb explosions, I would never question the technical feasibility. (Although we’re on the same page as to whether Pacer would ever be a commercial solution).
My point is that just as Pacer is technically feasible yet not cost-effective, scaled down H bombs detonated by lasers inside a reactor instead of under a mile of rock might also prove to be technically possible, but will never be relevant because it will always be far more expensive than easier methods of power generation (most likely including Pacer)!
Why are people still fascinated by the prospect of running yet another incredibly expensive science project that has already been run many times and which, as it “advances”, only proves that almost any alternative approach to power generation would be more practical?
Surely the mythology of fusion power is “limitless cheap power”. That is the false promise that I rail against. A paradise myth being held up as a hope and salvation that is not even needed, let alone practical.
daveza…
Yep! The fusion industrial complex has noticed how the climate scientists have used the purported climate crisis to get endless grant funding, tenure & fame from the gullible public, all courtesy of the eco-theologist masses and their willing politicians. Rent-seekers!
If funding slows down, just fire up the a 2 MJ laser to produce 3 MJ of fusion energy & hawk the finding over the MSM. But neglect to mention the laser needed ~ 300 MJ to produce a 2 MJ beam. The MSM will not notice the discrepancy.
Precisely right!
Was hunting around for the perfect Xmas present for the grandson and then I spotted it in top nick glowing in the back of the junk shop-
https://www.msn.com/en-au/news/other/santa-brought-me-uranium-the-most-dangerous-toys-in-history/ss-AA15ryTG
I want one :-). You can have a lot of fun with alpha particles, like using a powerful permanent magnet to bend the particle stream to jump over a cardboard shield and strike the detector – you have to line everything up just right.
Eric
This is something I heard about yesterday. Any thoughts on this:
https://www.safireproject.com/
Really nice web site, slick production values. Lots of keywords. Hope they come up with something. Meanwhile, I’m sticking with cold fusion.
Morton,
Here is a 2015 Michael Shermer piece on the Electric Universe group who are backing the SAFIRE Project. Shermer asks some good questions for which they apparently had no answers.
https://www.scientificamerican.com/article/the-difference-between-science-and-pseudoscience/
appreciate the reply. I had never heard of the EU group. Specifically, they claim to have had some success with their experiments, and they claim to have had other scientists confirm some of their findings.
Just wondering if there is something interesting happening here based on this pdf:
https://www.safireproject.com/science/ewExternalFiles/SAFIRE-Phase-Three.pdf
More fear mongering. It appears to be the popular dance of the day.
Would be interesting to show the length of each generation..in any units. Like to see Length on an XY grid or number of decimal places where change is happening.
It’s quite obvious. AI is being developed as the ultimate unarguable reason for ALL the bureaucratic cock ups ever made, here on in.
Ever tried arguing with an AI decision? If you don’t fit the pattern mate, you don’t exist mate.
Al…..not AI….but Al Gore sez just 7 more years…7 more years to fusion power and….an ice free arctic.
Seven years may be right, but his definition of a year is the time it takes our sun to revolve around the galactic center.
So about 7*230 million years or 1.6 billion years. Pretty close I reckon, but I don’t think old Sol will go red giant quite that soon.
“Of high programmatic importance are approaches that support the realization of a fusion pilot plant on a decadal timescale.”
Why? We are already there, having heard for decades that realization of fusion power is a few decades out. Same old same old.
For even higher “programmatic importance,” how about putting AI to work assimilating climate data, like the CERES 1 deg X 1 deg hourly values of LW emission and SW reflection? Task it to isolate non-natural from natural variation.
“ The Register; President Biden recently urged the DOE to produce viable commercial nuclear fusion by 2032. So the DOE is offering money to AI researchers, in the hope that the magic of AI can help make the magic of nuclear fusion a reality.”
So, are we talking SkyNet or The Matrix?
If they want fusion, I think Super Glue or JB Weld would be a lot cheaper.
There’s already been research in the field!
https://wattsupwiththat.com/2012/05/29/new-branch-of-science-coined-on-wuwt/
Not enough effort to actually accomplish anything but enough to claim credit for “inventing” fusion if the research turns out to be successful.
Programs to find the shortest lines connecting many points have been commercially available and used by the trucking industry for many decades.
For almost 50 years it has been know that mainline approaches to nuclear fusion– inertial and magnetic confinement– would NEVER lead to commercial fusion power plants. Yet the ‘crats never admit failure. All the AI in the world won’t change the Laws of Physics.
Fortunately the private sector has been investing their money in fusion alternatives and making noticeable headway. Aneutronic “Self-confinement” machines manipulate the plasma to fusion ignition temperatures and density.
Which you’ve been claiming here for years. Why not just give it a rest until you can come back and rub it in my face? Believe me, I would be ecstatic to be proven wrong.
Actually there is constant progress at LPPFusion. There are many steps towards achieving ignition and the team there is making progress. Much of it is currently on the engineering side where new components have been developed to achieve the goal.
Having achieved 2 of three criteria for achieving fusion convergence on the third density criteria is important news for those who appreciate developments in science and engineering. Especially since these developments are financed by private funds instead of massive waste on dead end approaches to fusion. Capisce?
See LPPFusion.com
Certo che capisco!
What I understand is that there are two tracks in the fusion rent-seeking scam.
You can make your career sucking down government funding from idiot bureaucrats and politicians or you can go for the gullible people with more money than brains in the private sector.
Now ovviamente if you’re going after the private sector suckers you need a story line where all the government initiatives are doomed to failure, but our mad geniuses have magic beans that will solve the problem in a totally new way.
Let us remember that your DoE paid for the birthing of PD Jones et al Univ. of East Anglia “global warming” onto the world.
I have just been running eye over hardcopy of TRO27 “A grid point surface air temperature data set for the Southern Hemisphere”.
Dated 1986 and there was an earlier TRO22 for the Northern Hem, I had always assumed the DoE paying big $s to Jones & co was birthed in the Carter years.
Just eyeballed the first 10 or so pages and am amazed that terms such as “urban heat islands”, “urbanisation” or “urbanization” never bothered the typesetter.
I thought we solved fusion last week?
Jamaica:
That made me chuckle! Their announcement was the equivalent to the “its worse than we thoght!” climate studies that always surface just before a big climate conference.
My favorite of those was Karl’s attempt to re-do the ARGO SST data in an attempt to rid the world of the ~1998-2015 “Hiatus” in warming.
Step back to the desire:
1) Keep fusion plasma stable.
A demand immediately complicated by government:
First a small idea, optimize lasers to everything needed to keep fusion plasma stable.
That desire immediately informs anyone with Federal experience that they have a committee listing of fusion plasma possible needs and they want a fool to work it out.
So they offer bribes for ideas, that the committee fools think will help them come up with new ideas. Mediocre will judge pathetic through mediocre ideas. Superior through outstanding perfection suggestions will be beyond their capabilities.
Beside, full employment is guaranteed until they accidentally use a genuine paradigm idea game changer.
That solution, they never want to discover, their salaries depend upon it.
The analogy is building socialism and the withering away of the state.