Audioburst Search

We Still Can't Explain Quantum Computing!

Automatic TRANSCRIPT

Good Morning Walk into marketing over coffee. I'm John Wall. Christopher pen and we are back we've got you back. It's funny if you haven't been traveling but it's just been weird with all kinds of stuff going on and we've had other interviews and the first one was. Ibm Think Twenty twenty so this was the first time it's all virtual. How did all that go? All things considered it wasn't bad. I still got to do the things I wanted to ask people the questions. I wanted to ask them. I know a lot of folks were like. It's not the same like yeah. Of course there's no you know terrible buffet line with You know a five hundred people in front of y'all all griping about the same hotel chicken. It was two days Live keynotes then a ton of on-demand sessions and the Nice thing is that because it's all all virtually begin with. They're leaving the on demand sessions up for thirty days so you can just zip on and take whatever classes and stuff you wanted all free which also is a nice change from the normally twenty seven hundred dollars a ticket. Yeah right twenty. Seven hundred ticket plus downtown San Francisco. You're spending five thousand dollars on your hotel just exactly. Yeah Whoa I remember when I was looking at the booking for this is like nine hundred bucks a night for a double bed in San Francisco like well. I do. Miss the serendipity of being able just like hang out and meet folks but I don't miss the bills. Well it was funny to you. Did you got picked up an article. That getting syndicated everywhere about doing this thing. Virtual and it was funny. I loved your phrase awkward for those lunches where they like four sixteen people around the table and have to train entertainer. Educate each other. When you don't even know each other and of course there's people at the table you like. I don't WanNa talk to you at all. Exactly the big things for me from the event though in terms of usable takeaways. There's a lot of really cool stuff. There's a better session this year on quantum and explaining what the heck quantum computing is. I still can't explain it. I can see the visualizations and I swear the visualizations look like you know the hollow maps that they had in star wars the death star's internal layout running for over a year. We've talked about this. How like even the people that created still can't explain it in a way that the rest of US humans can understand but at least did you get anything closer to as far as like you use cases or why. It's better to be able to explain it. I think of it as massive parallel computing. Where you have the ability to look at every possible permutation of a problem at the same time as opposed to having to do them one at a time. Which is you know how you solve problems now with computers. So for things. Like being able to simulate molecules and their interactions for pharmaceutical development. It's going to be a game changer. For that anything where you have to do a lot of computation in parallel all at once. It's going to wreak havoc on. Cryptography on encryption things because today's strongest encryption standards a quantum computer will be able to crack solvent about ten seconds. Because it can essentially you brute force test every possible combination of of a password all at the same time so that'll be interesting. The bigger thing for me personally was this concept. Be Ragam Sriram Renaissance name correct into idea of neuro symbolic artificial intelligence which is a blend of two different types of AI. So if we go back in history from the fifties to the eighties or nineties There's this concept of symbolic classical. Ai We would call these expert systems. Where you try and load. As much you know cleaned and curated data and hand-build allot of the if this then that kind of rules this type of a it doesn't scale well barely scales at all from the two thousands. Do Today we've had called neural a where we essentially dump trucks full of data into machine at a you learn from this and you figure it out and you write the code. The challenge neural. Ai Is that it's really easy to screw it up. There are the classic examples like painting white lines around the self driving car and then he doesn't know what to do. It just goes in circles or being able to enter like dating marketing automation system. And it just blows up your leads going algorithm so the concept that IBM's working with is the idea what if you use symbolic Ai? Which is the older version? That doesn't scales but as finely tuned and used as the guardrails for the neural a is so to say like these are the instead of trying to create one off exceptions. Like oh don't do that. It's like these are. The lanes are allowed to drive in in some cases literally and ideally get to a better overall than either system can do independently which I thought was really cool. The applications for example one the challenges we have in marketing is with today's AI. The ability for to generate language is not bad. But it's still not there yet like you read a piece generated purely by like something's not right here if you can imagine using symbolic Ai. Say like this is what I expect a blog. Post to look like these the rules essentially if a blog post and then the neural. Ai Fills in the gaps but keeps to the rules like okay. I can now create a blog post. That is intelligible that someone would actually want to

Coming up next