« Larry Johnson--Two Important Transcripts Shedding Light on the FBI and Two of Their Informants | Main | Flat Iron Steak tonight (inside) »

11 May 2020

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

BABAK MAKKINEJAD

This NIST achievement is solid and commendable but it is not going to bring Quantum Computing any closer; which - like Nuclear Fusion or String Theory - is another White Elephant of Physics - but is good for getting funding and getting promoted.

The fact is that any quantum computer - indeed any computer - requires the state of the computation to be copied at will.

Quantum states cannot be replicated, copied, or otherwise reproduced. This is a fundamental property of Quantum Mechanics.

All those researchers in the so-called Quantum Computing arena are aware of it and are also aware of the fact that they have no clue on how to address that issue.

So, like the proverbial drunk who has lost his key in some dark street corner but is looking for it under the street light since "that is where the light is", the researchers have gone elsewhere.

But, governments wish to find a way to decrypt other governments ciphers that are based on 50-digit prime numbers and the scientists are selling Quantum Computing to them.

By itself, this is not a bad thing, it keeps a lot of Ph.D.s employed and who knows, may be the Donkey could be taught to speak!


Walrus

“Any sufficiently advanced technology is indistinguishable from magic” - Arthur C. Clarke.

......And of the Eleven authors of the paper, two are Chinese of unspecified nationality and two have Indian names. What chance this technology stays in America?

stueeeeee

In our search to find the divine workings of our creator or to confirm that a cold random mutation spawned us, we will/have become Gods. This breakthrough will usher true AI...singularity. Computers that will be able to think recursively over and over and over to the point of consciousness. I sigh in awe and shiver at the thought of that moment.

The Twisted Genius

Impressive accomplishment, but that is just the hardware. IMO the real advances are being made on the software side. One Soviet trained AI cyberneticist I knew summed it up when he told me his goal was to create a machine that was capable of making a leap of faith. I taught myself Assembler and Prolog just to be able to keep up with his conversations. It made for some interesting debriefings and taskings.

A good friend of mine developed a new branch of geometric algebra in order to write his own AI programming language. I noticed some Prolog III code as it scrolled by in my first meeting with him so he assumed I knew of what he was developing. I assured him I was no more than a dabbling amateur, but I could appreciate his work. We worked together for a few years. Some of his early ideas and guidance were incorporated into IBM's Big Blue technologies. His "machine" can run on minimal hardware rather than relying on supercomputers. It does deep analysis and learning like a lot of current AIs, but is also adept at myriad natural language applications. It still doesn't make that leap of faith, as far as I can tell, but it seems to be heading that way. I like to think my contribution to all this was to impress on my friend the importance of not trusting the government, any government.

Fred

Walrus beat me to the question, "what chance this technology stays in America". It is probably gone already, so hopefully, as Babak points out, there might be no there, there.

Dave Schuler

Practical quantum computing has been just around the corner since I was in grad school and that was nearly 50 years ago. As suggested above, like practical fusion reactors, quantum computing is likely to remain just around the corner for the foreseeable future. Although there were some interesting developments along those lines recently just a few days ago!

JP Billen

I thought the Germans had already done this several years ago. What is new and different about the NIST transistors? Is the process simpler and faster? Or is it the stability mentioned in the scitech article?

Barbara Ann

With unparalleled computing power comes unparalleled responsibility. As I write this, humans are still deemed to possess the greatest known intelligence, by most measures. I have to wonder how much longer that will be true, with the headlong race everywhere to develop AI.

Some very smart people are warning of the dangers of us one day creating an intelligence superior to our own. Henry Kissinger is one. Here is his 2018 piece in the Atlantic on the subject, entitled "How the Enlightenment Ends".

https://www.theatlantic.com/magazine/archive/2018/06/henry-kissinger-ai-could-mean-the-end-of-human-history/559124/

Another great thinker; James Lovelock, has cautioned against the development of AI, which he describes as a new form of life. He focuses not so much the intelligence aspect, but on the speed of 'thought', as electrical impulses travel around 1,000,000 times faster in a computer than in a human brain. He says intelligent computers talking with us will be like us talking with a tree.

I'm all for developing better tools, but the quest for super human AI seems to me the ultimate Promethean project. We should be very careful what we wish for.

blue peacock

Barbara Ann,

We are generations away from the AI that James Lovelock fears. Current AI is rather prosaic. Training software algorithms (neural networks) with patterns. For example training it to distinguish between images of cats and dogs.

BABAK MAKKINEJAD

Barbara Ann

The purveyor's of AI, just like those of some form of Androids, are men who dislike women.

They dislike women because they cannot conceive children and bring them into the world.

And they dislike women because they could not score when younger.

They want to make women obsolete.

Once they have succeeded in doing that, they will proceed to make man obsolete as well.

If I were a religious man, I would have stated unequivocally that they have been listening and have been successfully tempted by Satan.

Personally, I think the application of CRISPR techniques have a much better chance of making women obsolete, by creating chimeras that are sex toys for men.

Robert

There is a new TV show on Hulu called Devs. It is set a few years in to the future, and explores a secret Silicon Valley Quantum computing lab run by an enigmatic CEO.
The writer/director Alex Garland (Ex Machina) spent time at Googles Quantum computing lab in Goleta CA among other places to get a handle on the Science. Highly recommend it.

walrus

If you haven’t read this, you should really enjoy this; The 1959 lecture by Richard Feynman that established this entire field: “There’s Plenty Of Room At The Bottom”.

http://www.phy.pku.edu.cn/~qhcao/resources/class/QM/Feynman%27s-Talk.pdf

walrus

Link to an American copy of the talk:

http://calteches.library.caltech.edu/1976/1/1960Bottom.pdf


It is ominous that the first google entry - that I linked to, is Chinese.

Vegetius

I believe we are going to need a Butlerian Jihad sooner rather than later.

Just some guy

In the semiconductor industry highest end products are made to 70 Angstrom design-rules; where customers bitch about oxide thicknesses being out of spec by Angstrom or so..

A hydrogen atom is about 0.74 Angstroms across. A silicon atom is roughly 2.37 Angstroms. The problem with single atom devices is that to get the signal out, the supporting structures are much larger, so the effective size of the device is not an atom. Secondly, on this scale the results are probabilistic, so there is s certain noise floor you cannot beat. A transistor is an amplifier of current, so scaling down to an atom as the core of device has to provide some advantage over over a more conventional structures that make it worth it do deal with signal to noise ratios and fabrication issues. Faster switching speed (“on” to “off”) and less dissipation of waste energy are two properties that might make it worth it to scale down.

The article states the device would lead to making Qubit devices, but that is not what was made. They made transistors. For a manufacture-able process, they will have to get away from using a scanning tunneling microscope and replace it with something else. Additional circuitry will be needed to put sets of them into a controlled superposition of states required to make a Qubit device.

Its very nice work by NIST, but somewhat breathlessly miss-represented by Scitechdaily. Devices that take advantage of quantum tunneling have been around since early days of semiconductor industry. What has been happening is a ruthless scaling down. The part about quantum mechanical entanglement should be read strictly as entertainment. The time-evolution of isolated quantum mechanical systems is probabilistic, but follows rigorously defined math that cannot be boiled down to an understandable pop-science article.

We live in an age of miracles and wonders.

The Twisted Genius

Bakak,

Your ideas about the purveyors of AI are way off. All the AI developers I know either are married, most with kids or still single and have no problem getting all the strange their little hearts desire. I will say they are a quirky and odd lot, but they certainly don't hate women, seek to replace them or waste their time listening to Satan.

Mark Logan

Barbra Ann,

Turing was a remarkable fellow. He not only created the first useful computers he created the ultimate test for AI: The Turing Test. Even with those crude computers of his day he could tell where it was going,

No computer will become indistinguishable from a human until it has motivations of it's own. They will remain machines until then.

Babak Makkinejad

Twisted Genius:

Just because someone is married to one does not mean that he like women.

ked

“ researchers must find a way to make many copies of these notoriously difficult-to-fabricate components.”

It will be relatively easy for them to get more $ out of NIST or DARPA... more challenging to get numerous industry players (& gov) lined up to sponsor production / repeatability development. Maybe the PR blitz will help. So in 3-5 yrs one might be able to estimate if it’s 5 more yrs, 10, 20 or never. At least there’s a shot (10%?) of something useful (if unrelated to the original goal) spinning out of the core effort.

One of the many issues w/ these kinda efforts nowadays is the intense concentration / duration / tiny-scale / control of power required to keep bleeding edge processes launched & reliably operating beyond a lab proof-of-concept. All part of the fun.

The comments to this entry are closed.

My Photo

May 2020

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30
31            
Blog powered by Typepad