E598: Babak Hodjat, Sentient.ai Co-Founder & Chief Scientist (& main inventor of tech behind Siri), is creating artificial intelligence that learns and adapts with our every decision

watch

listen

share

Share on FacebookTweet about this on TwitterEmail to someone

about this episode

Babak Hodjat is Co-founder and Chief Scientist of Sentient, the most-funded artificial intelligence startup in the world, which emerged from stealth in 2014 after raising more than $100m Series C (the largest round ever by an AI startup). Though already going strong in financial trading (with 1.5+ CTUs globally), last week they launched Sentient Aware for E-commerce, which is already being used by the Canadian shoes.com and which Babak demonstrates for us. Babak is also the main inventor of the tech behind Siri, and he and Jason discuss the actual intelligence level of Siri, along with many other fascinating aspects of AI, including the ethics behind it, where it falls on the “worry” scale between simple memory chips and nuclear weapons, its potential impact on warfare, gambling, and self-driving cars, and the inevitable 20-year sushi bet! AI is capable of improving ourselves (in the case of health), knowing what we want before it even exists (in the case of e-commerce), but will it be capable of destroying humanity?

Never miss an episode!

Subscribe in iTunes: Audio (http://bit.ly/TwiStA) || Video (http://bit.ly/TwiStV)

Find us on Spotify Now.

Thanks to our sponsors. Click here to tweet your support (can edit before sending): http://ctt.ec/Gob7b

==============

Follow on Twitter:

@babakatwork
@sentientdai
@twiStartups
@jason


Launch Festival: http://launchfestival.com
Special thanks to the members of the TWiST Backchannel!
Support TWiST and join the Backchannel: https://recurrency.us/members/twistartups

 

0:00 – 3:19 Babak Hodjat invented the tech behind Siri, this episode will get people investing in AI

3:20 – 4:25 “The most complex we could think of was trading.”

4:26 – 6:58 “Independent problem solving… inspired by natural intelligence.”

6:59 – 9:52 “It’s the soft stuff that makes it feel human and warm and fuzzy.”

9:53 – 11:23 “We decided, let’s look at other technologies that work just as well.”

11:24 – 13:34 “It’s about decision making, it’s not simply about pattern matching.”

13:35 – 15:55 “We’re running on more than 1.5 million CPUs, running for months.”

18:20 – 20:40 “We used to give them names… we don’t anymore.”

20:41 – 22:10 “The system actually made use of the manner by which we execute… to take a position early on.”

22:11 – 24:42 “My traders have to go through millions of decisions, some of them artificially… and they have to come out of it not breaking the system.”

24:43 – 28:43 “AI today is nowhere near [causing mass hysteria].”

28:44 – 31:06 “Machines becoming sentient is still a few generations away.”

31:07 – 34:05 “DeepMind made some very interesting strides in applying deep learning to learning… Atari games from scratch.”

37:48 – 42:11 “This whole visual intelligence interface… was generated automatically using a combination of deep learning and online learning.”

42:12 – 45:04 “We started off with tags that were human-generated… automatically identify and abstract features… from the images.”

45:05 – 46:14 “Imagine that some day through your interaction with the system… figures out the shoe that is the best suited, and that shoe might not actually exist.”

46:15 – 48:26 “If you combine it with collaborative filtering, you can put people into buckets.”

48:27 – 49:40 “If you actually have distributed AI running on the point where the data… is being acted upon.. there’s no need to store it.”

49:41 – 52:56 “No two people have the same reactions… we can discover things that are very very personalized.”

52:57 – 54:34 “You can test the hypothesis within the individual’s own dataset.”

54:35 – 56:58 “What is the politics of getting a PhD?”

56:59 – 58:28 “A lot of us AI practitioners have already seen the power of AI… we want the rest of the world to see that.”

58:29 – 59:40 “There were some physical limitations that were holding the industry back.”

59:41 – 1:01:23 “We make use of third-party data centers… designed for peak capacity.”

1:01:24 – 1:03:35 “Personal assistance… is farther along… we never expected to do speech recognition at this level of accuracy.”

1:03:36 – 1:05:55 “I think more and more of our decision making is going to be done by distributed AI systems.”

1:05:56 – 1:07:51 “It’s keying off of tags it automatically identifies… the concept of a buckle, per say, may not be a concept the AI has identified as a differentiator.”

1:07:52 – 1:09:45 “The number of years before a Michelin starred restaurant uses computers to set the menu.”

1:09:46 – 1:11:01 “There’s a list of recs that we have up there, we’d love to have [developers] work with us.”

1:11:02 – 1:13:42 “This technology, you could literally make a drone… that figured out how to kill people faster.”

1:13:43 – 1:15:52 “If you made a computer that was trying to beat other people, especially if it was covertly doing so, then that would be kind of unethical.”

1:15:53 – 1:18:00 “If all these cars are acting in a reasonable fashion, the chances of these outliers are going to go away.”

1:18:01 – 1:20:45 “I’m very hopeful for the future of Iran.”

1:20:46 – 1:25:42 “It takes several generations to rebuild those countries… and bring them back to the present day.”

more episodes