I was deeply delighted recently when one of you raised the hope that WAAS, the World Academy of Arts and Sciences (WAAS), might take over the essential and urgent task of bridging the gap between serious, advanced RD&D and the research mission of the new UN agency on AI.
That was a great hope, like the hope I felt when Kerry, Guterres and EU announced they would focus new efforts in trying to reduce the threat of human extinction by the most dangerous type of climate change. Unfortunately, good intentions sometimes cause more harm than good when the implementation does not account for the full range of science, technology and rational market design needed to cope with very difficult new challenges, requiring even a new scientifically grounded understanding of human minds. The new challenges coming fast in the global internet require many massive changes, as fast as it is possible to do right; if we do not work VERY hard to start mobilizing the best that the old NSF and IEEE have ever done, a new UN agency can easily become prematurely ossified in many ways.
** IN THIS POST** ... I will address just ONE crucial aspect, the need to openly develop the most advanced new quantum information technology, ONE of the important key ingredients of the new US act in this area:
https://www.aip.org/fyi/2022/chips-and-science-act-enshrines-policy-new-nsf-technology-directorate
DOES WAAS know what it needs to know to fill his gap for the new UN agency? Not without special help from SOMEONE. Through this post, I will try to help that someone -- and pray that one of you chooses to be that someone -- but all I can do is flag some key points about quantum information crucial to the choices before us now, radically different from what you may know as yet.
** IN THIS POST** ... I will address just ONE crucial aspect, the need to openly develop the most advanced new quantum information technology, ONE of the important key ingredients of the new US act in this area:
https://www.aip.org/fyi/2022/chips-and-science-act-enshrines-policy-new-nsf-technology-directorate
DOES WAAS know what it needs to know to fill his gap for the new UN agency? Not without special help from SOMEONE. Through this post, I will try to help that someone -- and pray that one of you chooses to be that someone -- but all I can do is flag some key points about quantum information crucial to the choices before us now, radically different from what you may know as yet.
WHY is new RD&D in quantum information S&T (QuIST) so crucial and so urgent as the Chips Act rightly highlights, thanks in part to Senator Warner of the Senate Intelligence Committee??
Two reasons:
(1) New cybersecurity threats, on everything from global monetary systems to fleets of weapons more advanced than you know about, and electric power, are more important and urgent than anything else being debated now for "the new internet." The new agency needs to START there, including the proposal a few years ago from Xi Jin Ping for international collaboration to limit or prevent the new kinds of backdoors in software and hardware which utility executives on IEEE USA EPC explained a few years ago (see attachment).
(2) True Quantum AGI, as defined in my WCCI award acceptance speech last year (abstract attached), is ONE of the two new directions rightly stressed in the IEEE EPC statement, making it possible to detect and prevent the kinds of hardware backdoor which have already started to cause major problems and are on track for far worse.
But should you believe that? Why, and how?
Just this morning, my wife rightly cited Gershenfeld of MIT who tells us not to overestimate the importance of modern quantum technology, for which he was one of the very most important real pioneers. The vast majority of people now being funded for quantum computing in US and EU are working on Quantum Turing Machines (QTM). They owe a lot to Gershenfeld, and to David Deutsch of Oxford (who invented the Quantum Turing Machine). Just as EU climate action tends to channel certain established energy and political causes, WAAS will be under huge pressure to listen on what Gershenfeld just said, based on a very deep understandinding of QTMs.
But from 2007 to 2015, I had the unique good fortune of leading a unique funding activity at NSF
https://arxiv.org/abs/0709.
https://photos.app.goo.gl/
The new agency NEEDS to be fully informed of this entire territory, including emerging new options (being pursued outside the US and EU more effectively than here right now).
THE QTM HAS LIMITED RELEVANCE TO CYBERSECURITY, FOR EXAMPLE, but the Quantum RLADP technology
(the core of QAGI) is a very different story. THIS REQUIRES THAT A NEW AGENCY FULLY UNDERSTAND THE FULL NEW QUANTUM FOUNDATIONS ALREADY ADVANCED BY DEUTSCH AND GERSHENFELD (WHOM I WOULD LOVE TO HELP ON THIS), applied in a totally new way beyond what QTMs can do.
One of the big problems I learned in running QMHP (and serving as NSF member of the interagency Quantum Information S&T Coordinating group, QISCOG) is that MOST "experts" submitting proposals and plans had fatal errors in their understanding of David Deutsch's physics and how to apply it. (The physics is much more universal than the QTM example.) Thus I thank Yeshua ben David for arranging a mid-level seminar explaining the most important points
which key people in the new agency need to know about, to take the lead in new relevant quantum technology:
What can Quantum RLADP do orders of magnitude better than the existing (inadequate) cybersecurity hardware technologies like bromium or even bromium enhanced by DWave quantum optimization?
Effective parallel quantum search.
In practical terms... it is uniquely suited to solve "needle in a haystack" or "minefield" optimization problems, problems which combine the two. In technical terms, it offers orders of magnitude more performance for any number of flux gates or quantum dots (the most promising hardware embodiment) , when the value function or error function are highly nonmonotonic functions of the parameters to be tuned. This can be applied NOT ONLY to value functions or error functions defined by subroutines or circuits WITHIN a DWave kind of computing system, BUT ALSO TO PHYSICAL PLANTS TO BE STABILIZED WITHIN PHYSICAL BOUNDS, like the big laser I mentioned in my previous post. I call this extension "Quantum Annealing of Things," QuATh.
Please forgive me a confession. The full technical details of QuATh, including a workplan for phased implementation, are included in a pending patent disclosure.
In previous decades, I had information rights to backpropagation (a necessary enabler of the "deep learning revolution", one of the two which won me the IEEE Frank Rosenblatt award). Half our family income traces back to reassignment of patent rights of the most powerful CLASSICAL RLADP optimization technology. I obtained these rights only after strong encouragement from Charles Brown, then at the NSF Office of General Counsel. If the lawyers at that time had blocked any development or use of backpropagation until my retirement from NSF in 2015... I hope that we can find a way forward despite such realities, which are even more essential now than they were then (because of what I know in my piece of the puzzle, which I hope pass on as soon as possible).
Best of luck to us all. Given the precarious state of humanity in this century, we need a LOT of that luck...
Paul
On Thu, Jun 1, 2023 at 10:19 AM Paul Werbos <paubos@gmail.com> wrote:
On your nuclear question...which has new urgency this week because of the Global Digital Compact (GDC) which the UN Secretary General Plans to Unveil in about three days...I was very excited this week to hear the LLNL international presentation on the technical details of their work on inertial fusion at NIF.James Gaffney gave the kind of true cutting edge technical reality which we in IEEE have a unique ability to appreciate and understand.Unless you count stuff like cold fusion, I believe that LLNL leads the world in general in getting to ignition point.They got there with the usual DT fusion, but Perkins of LLNL previously led development of more advanced D-D fuel pellets, which would truly revolutionize our situation both on earth and in space. Better performance in the D-D work they are already succeeding with would get us closer to great economics, and to the far more benign D-D option. (DD creates about 10% as many neutrons as DT does, and has plentiful fuel in vast cheap supply.)Most exciting... and why I cc our CIS (outgoing) President ... Jim Gaffney, representing the two groups at LLNLmost crucial to the international community, stated that THE KEY TECHNICAL CHALLENGE NOW IS BETTER INTELLIGENT CONTROL (accuracy, stability, duration)!! This is a core IEEE CIS technology, where we have new options far more powerful than anything LLNL knows about YET (even though they outperform the usual machine learning mass tools, and have important tools already on github/LLNL). I suddenly wish my senator (Warner) and Congressman (Beyer) knew about the exciting new opportunity to build a new crosscutting team somehow to connect Gaffney's efforts out of LLNL with the newest and most powerful intelligent control tools available from CIS and other related IEEE communities. (For those outside IEEE, I attach the one page abstract I used last year to go with the Frank Rosenblatt award, with links to key aspects -- such as technologies which have been widely developed in China but not so much in US since my retirement from NSF in 2015.)However: this is not just about energy, climate and space.Thanks to a great presentation yesterday from the new Technology Directorate (TIP) of NSF,I hope and pray that we and NSF could connect with the Secretary General enough to see the great synergy between TIP and GDC. The GDC (and human security) will require a higher level of international cooperation, AND YET deeper technical competence, than the world has ever seen as yet. If there is any hope of getting that started, before the usual entropy sets in, it calls for strong and urgent efforts to build these bridges, so that the world as a whole can start to get and support what it needs from NSF and IEEE in greater abundance.I was stunned by the synergy between Threads A, B and C of the TIP presentation and the areas we have been moving forward on an international basis for the past few years...Best of luck to us all. (We need it!)
No comments:
Post a Comment