Google's quantum supremacy formally published; IBM disputes Google's claim of quantum supremacy and their use of the phrase; A power-law guides how much detail the brain saves; An argument that the danger of AI is that it will do what it's told; and a Steem essay argues for open access in scientific publishing
Straight from my RSS feed
Whatever gets my attention
Links and micro-summaries from my 1000+ daily headlines. I filter them so you don't have to.
pixabay license: source.
- Google scientists have claimed a massive breakthrough in cutting-edge computing with 'quantum supremacy' - At long last, the paper has been officially published, in Nature. As previously reported, Google solved a problem on a 53 qubit machine that would have been impossible under a classical architecture. Perhaps more significant than solving this highly customized problem, they also published a rule of thumb that is informally known as Neven's Law, saying "We expect that their (quantum computers') computational power will continue to grow at a double-exponential rate: the classical cost of simulating a quantum circuit increases exponentially with computational volume, and hardware improvements will probably follow a quantum-processor equivalent of Moore’s law, doubling this computational volume every few years." Another point that has been overlooked is that this may be the first example of computation that doesn't comply with the Church–Turing thesis, because it can't be simulated with a Turing Machine.
- Google: We've achieved quantum supremacy! IBM: Nope. And stop using that word, please - In their paper on achieving quantum supremacy, Google claims to have solved a problem that can't be solved on a classical computer (in a reasonable time frame). IBM responded that they know how to solve it in 2 1/2 days on a classical computer, and "with far greater fidelity". IBM goes on to say that 2 1/2 days is an upper boundary, and with refinement, they believe it can be done faster. Despite the dispute whether it's a true demonstration of quantum supremacy, IBM did acknowledge that Google made a major advance in the quantum computing landscape. IBM also detailed two reasons for avoiding the phrase quantum supremacy. First, it is misleading, because there will be always be some tasks that classical computers do better; and second, it evokes imagery of the phrase white supremacy. (Personally, I don't find the second reason to be persuasive... is the word, "supremacy" to be outlawed now? It seems paternalistic and belittling to me to think that the general public can't distinguish the meaning of "supremacy" in two completely independent contexts.)
- A Power Law Keeps the Brain’s Perceptions Balanced - According to a June paper, in Nature, the brain strikes a balance between encoding everything it can and remaining flexible to respond to noise in the sensory inputs. Research over recent decades has found that the brain encodes information in simple patterns, which led to the belief that the brain normally discards most of the input that's presented to it. However, later research found that those results were the consequence of flawed experimental designs. The experiments presented animals with simple inputs, so the encoding had to be done in a simple pattern. Now, in their new experiment, Carsen Stringer, Kenneth Harris, and colleagues, presented mice with over 3,000 images, and devised a new technique to record tens of thousands of neurons at the same time. In contrast with earlier studies, they found a higher-dimensional encoding. A particularly interesting aspect of what they found is that all dimensions were not encoded equally. A few dimensions captured most of the stimuli, whereas other dimensions filtered a lot f the stimuli out. Adding new dimensions, therefore, offers diminishing returns. They also found that the rate of decay for the filtering followed a power-law. Without reading the details in the Nature paper, it sounds a lot like Principle Component Analysis. The team has not yet identified the biological mechanism that causes the brain to use this encoding scheme.
- The danger of AI is weirder than you think - In this TED talk, AI Researcher Janelle Shane talks about artificial intelligence. With a series of comical examples, she demonstrates the reasoning behind her opinion that the danger of AI isn't that it will rebel against us. Observing that working with AI is less like working with a human than it is like working with "a force of nature" Instead, the danger is that it will do exactly what we tell it, in a harmful way that no one anticipated.
- STEEM My full comments for Inside Higher Ed's article "Where Research Meets Profits" - In this post, @dhimmel responds to a journalist about a recent controversy where, apparently, a professor received a take-down notice from a journal for posting his own article on his web site. @dhimmel notes that this is a consequence of the closed access journal system, and that the solution is to publish in open access journals. The post goes on to argue that another problem is the unpaid and anonymous nature of review, saying that journals should find a way to pay reviewers, and a quality review should be worth somewhere around $500, but opines that getting away from "toll access journals" is the most important problem facing science at the moment. In conclusion, the post argues for two immediate steps from journals: (i) Publish under an open license; and (ii) Publish peer review reports alongside the paper. (A 10% beneficiary setting has been applied to this post for @dhimmel.)
In order to help make Steem the go to place for timely information on diverse topics, I invite you to discuss any of these links in the comments and/or your own response post.
- Burn Steem/SBD - @null - 5%
- Cited author(s) - @dhimmel - 10%
- Fundraising for the Rustin Golden Knights Marching Band - @rgkmb-unofficial - 10%
- Posting and/or scheduling service (steempeak.com) - @steempeak - 5%
- Science, Technology, Engineering, and Math (STEM) curation on Steem - @steemstem - 5%
- Steem/API services (anyx.io) - anyx - 5%
- Steem/RSS services (steemrss.com) - torrey.blog - 5%
Sharing a link does not imply endorsement or agreement, and I receive no incentives for sharing from any of the content creators.
Thanks to SteemRSS from philipkoon, doriitamar, and torrey.blog for the Steem RSS feeds!