Isn't it astonishing that Microsoft has cooperated with a US national laboratory to use artificial intelligence to sift through 32 million materials in 80 hours, and have found 18 potential new battery materials, including solid-state electrolytes, without the use of quantum computing? Scientists have also been rewarded with the treasure and have begun to synthesize these materials and test them intensively, hoping to develop exciting new battery materials.
But that's just the surface of the message, and if you understand the logic behind it, you'll probably be even more surprised, as it's likely to signal an upcoming explosion of the human tech tree, and it's likely to be ferocious beyond your imagination!
The future of the big killer
The hidden information behind this news is that artificial intelligence has actually found not 18 kinds of battery materials, but 23 kinds of battery materials, of which 5 are known, that is to say, they have been discovered by humans through scientific research, do you think of the logic behind it? So many scientists have been researching for so many years, so much money has been invested, and 5 kinds of definite battery materials have been found, and artificial intelligence has found 23 kinds of 32 million kinds of them in 80 hours, almost effortlessly, if the test results show that those 18 are available, then how much impact will this system have on the development of human science and technology when used for similar other screenings? Can you imagine?
This system is the new Azure quantum element launched this year, Microsoft's future exploration of the microscopic world, Microsoft's cloud service built to accelerate scientific discovery. It leverages the latest technologies in high-performance computing, artificial intelligence, and quantum computing to help scientists design, code, and present molecular models to simulate and predict problems in chemistry and materials science through tools such as Copilot.
In layman's terms, this system uses the advantages of quantum computers to simulate the occurrence of chemical reactions at the molecular level to find the various new materials we need to solve various problems. But now quantum computers have not yet been born, so this system is simulated by high-performance computing and artificial intelligence, and even then, it has exerted great power and achieved incredible results.
Microsoft's own statement is that some of the customers who use this system to find materials have shortened the time to obtain results from six months to one week, some have expanded the number of candidates screened at the same time from thousands to tens of millions, and some have increased the speed of chemical simulation by a factor of 500,000, which is equivalent to compressing a year's research time to one minute. Previously, on Twitter, the CEO of Microsoft also claimed that it is necessary to compress the chemistry and materials science of the next 250 years to the next 25 years, which is to make human beings immortal within 25 years.
In principle, our body is an extremely delicate machine that uses chemical energy, and it will eventually die because of the wear and aging of various parts, but if we can find various symptomatic compounds, of course, we can completely solve the problem of wear and aging of the machine.
Discover new materials
Microsoft is partnering with Pacific Northwest National Laboratory (PNNL) in the United States.
In the past, to find new battery materials, scientists needed to go through complicated papers, find various candidate material combinations to study, and then conduct experimental screening, most of the time in vain, so some people say that the so-called materials science is luck, you may be the Nobel Prize, if you don't, you can only be three feet high. The most speechless thing is that these failures will not appear in the paper most of the time, which leads to everyone repeating the failed screening again and again, wasting time, energy and money.
This time, PNNL researchers worked with Microsoft to list 32 million possible materials using artificial intelligence to develop new battery properties according to their needs, and then filtered out chemically unstable ones through algorithms, first reducing the list to 500,000 and then to 800. Then, using a combination of artificial intelligence and high-performance computing, molecular dynamics calculations were performed on these materials, simulating the movement of atoms within the materials, and more accurately analyzing the energy states of the candidate materials, again narrowing the list to 150. Finally, high-performance computing was used to evaluate the availability and cost of materials, and finally 23 materials were selected, and the whole process took a total of 80 hours.
Scientists were pleasantly surprised to find that one of these materials, a solid-state electrolyte that is more stable and cost-effective than the materials currently in use, may be because it contains sodium and lithium, which had previously been ruthlessly ignored because the combination was often considered harmful to batteries. However, this screening found that the two seem to help each other, but will reduce the amount of lithium used by 70%, so it will bring a huge cost advantage. Scientists have synthesized these 18 materials and are testing them.
Quantum Artificial Intelligence
I don't know the results of the test yet, but that's not the point, the point is that AI may indeed be able to significantly accelerate scientific progress. Previously, scientists believed that with the development of hundreds of computers with real qubits, the complex structure of the penicillin molecule could be simulated, and a few thousand could crack all the current encrypted data. Even if it can't be simulated with a quantum computer, practical screening has been achieved through algorithms and artificial intelligence, and most importantly, it's really unimaginable that it's so fast.
And institutions such as IBM and Harvard University have made one breakthrough after another in the manufacture of general-purpose quantum computers, which may eventually be born within 10 years. Microsoft also claimed in 2022 that it had achieved its first milestone on the road to a quantum computer: the creation and control of the Majorana quasiparticle, which is believed to allow them to design reliable logic qubits to create a truly super-quantum computer.
It is in this context that Azure quantum elements may have been born, with the ultimate goal of using artificial intelligence, which can understand human language, and quantum supercomputers, which can understand nature's fundamental language, chemistry, and the combination of the two can enable quantum chemistry to simulate and analyze chemical processes in the microscopic world with unprecedented scale, speed, and precision, leading to tremendous advances in chemistry and materials science. So the whole system, we might as well understand it as quantum artificial intelligence, well, it doesn't seem to have anything to do with human beings, do you already feel a little chill down your spine?
A double-edged sword
For example, the use of AI screening to develop drugs has been a relatively common phenomenon, and there have been many successful cases. Like a miracle drug during the pandemic, Pfizer's Paxlovid was developed with the help of supercomputer AI algorithms. The University of Toronto, Canada, used an artificial intelligence drug discovery platform this year to discover a new target for hepatocellular carcinoma and develop a potential therapeutic drug within 30 days. According to Exscientia, a British AI pharmaceutical company, AI can reduce drug development time from the previous average of four and a half years to about 13.7 months, a full three-quarter shortening!
However, great power also means great risks, such as quantum artificial intelligence can go to the quantum world and simulate the screening of drugs and materials at the molecular level, and of course, it can also screen out poisons and more powerful explosives. One of the most famous examples is that in September 2021, two scientists used the artificial intelligence platform MegaSyn to generate 40,000 deadly molecules with toxicity equivalent to VX in less than 6 hours, and VX nerve agents only need 6~10mg, that is, particles the size of a few salt grains, to kill people.
Of course, these are all within our control as humans, and even if there are bad guys, it is unlikely to pose a large-scale risk. But in the future, artificial intelligence is likely to be ubiquitous, and with the blessing of quantum computers, and then AI develops self-awareness, who knows what will happen at that time? We can't go to the microscopic quantum world, they can go in our place, and who knows what they will do there?
So I always feel that with AI and quantum computers, we humans seem to be superfluous in this world, they can do everything, what is the meaning of the superfluous things? Fortunately, we are just getting started, and we can enjoy the conveniences they bring, such as the development of a new drug to cure diseases and restore our health, the discovery of a new solid-state electrolyte that can charge an electric car for 10 minutes and give it a range of thousands of kilometers, such as the snap, which allows you to score 100 points on every test, and you don't have to use your brain anymore.
Well, it can't be snapped anymore, and we'll be really redundant if it snaps again. Actually, if you find it, many people are already redundant now, but this does not prevent our world from functioning harmoniously, so what is the need to worry about it? The explosion of science and technology is coming, so let's make it more violent.