Have you ever thought about what happens when digital AI systems are put into a physical context?
Giving artificial intelligence carte blanche to operate in any area of the real world gives it additional dimension, makes it more woven into our lives, in profound ways.
For example, think about some of these ideas as presented by Sertac Karaman, where he talks about some of the work going on at the MIT LIDS lab…..
“It’s the kind of laboratory that really applies AI in the real world, where the rubber hits the road,” he told students, in describing the work that LIDS does. “And so we start to think a lot about: how would AI get out of the digital ecosystem, the digital environment, and then go out and connect to … physical or social environments?”
These questions, he said, drive his passion for AI exploration.
“When AI hits the physical world,” he said, “you’re not just moving bits around, but you’re also moving atoms, molecules. I’m also very excited when (AI) hits the social world. You’re not just moving bits around, but you’re moving beliefs…at massive scales!”
He gave the example of social media, which, of course, many of us think of first, given all of the headlines around how our biggest platforms affect our behaviors.
Anyway, Sertac also presented us with three examples of key figures on the IT landscape that inform his work: John W. M. Bush, Claude Shannon and Gordon Brown.
“He is known for literally inventing the computer,” he said of Bush’s work starting in the 1930s. “When you’re doing that in 1930, … there’s no digitization, you build a computer that’s… a big mechanical system.”
Later, he noted, Shannon was instrumental in moving the needle toward what we’re into today. “Five people got together in Montreal in 1961,” he said, “One of those five people is Claude Shannon – he wrote a paper at that time, about (something) called machine learning. He advocated …(getting rid of) these model-based methods and rule-based methods just collect(ing) data, and teach(ing) systems how to learn.”
As for Brown, Sertac noted his work on defense, in what was a vitriolic time.
Moving on, he tied this earlier work into what we’re looking at now with AI and automation, describing what we may soon experience.
“It will be the very first time that in humanity, we experience a single system that can talk to a million people at the same time,” he said. “It’ll be talking to one of you, and then another one, and you will understand that in the background, you guys are actually talking to each other – we will know, given your responses.”
He talked about the power of information to transform beliefs, pointing out, for example, financial use cases. Then he went over some current priorities in research.
“Our current focus areas are very different than in 1940,” he said. “The big major focus area is machine learning, we’ve continued that throughout, … we have a good kind of a center of mass of machine learning, we probably have about half the faculty working on machine learning at MIT.”
Another focus area he went into is really, at the end of the day, closely related to game theory. I thought it was interesting when Sertac described a computer program that can play a game called Diplomacy, which is based on very human responses, creative and strategic alliances, things that a few years ago, you would have supposed computers to be bad at!
“(It) tricks people,” he said, describing such an intelligent engine. “(It) says: ‘hey, you know, I think you and I should make an alliance, and go attack this person. Because I talked to this other person’ … and it never talked to that other person. It’s clearly lying.”
Creepy or cool? Or both?
He also discussed reinforcement learning, and applications to autonomous systems. Specifically, self-driving vehicles, where he has actual executive experience.
“So I … worked on self-driving cars for decades,” he said. “There was an event called the DARPA Urban Challenge that kicked off the entire industry. So as a part of that, I was one of the 10 graduate students who programmed MIT’s (entry).”
Then, later, he got involved on the business side.
“I’ll tell you a little bit about my second company,” he said. “We were five co-founders, all from MIT, we covered the whole ground. So I’m from LIDS, we had another co-founder from CSAIL, we had one person who was a Sloan MBA …We had one more person who was from the Media Lab, and we had one other person who’s actually kind of Media Lab Plus Harvard, … We started the company in 2015, people were saying that ‘next year in 2016, all cars will be autonomous… it’s just going to happen right away.’ We thought, looking at technology, … that’s just impossible. We thought: ‘maybe we can take 10 years, and do another product that would be different than fully autonomous driving in all conditions’…”
Eventually, he said, they started looking at the use of shuttles, infamous for their inconvenience.
“(We thought maybe) we can use autonomy technology, to take shuttles from something everybody hates, to something that everybody loves,” he said.
“I think we’re going to look back at this moment, millennia later, and think that this was a huge moment, that we would finally make some artificial life form that will do the same things, that differentiates us from everybody else. It’s huge.”
He urged students to be excited about the underlying phenomenon, not the individual use cases, so that they’re not just running after the next shiniest thing.
As examples of guiding project vectors, he mentioned aggressive personalization, automated data engineering, and digital worlds, as well as front-running examples of interactive robots.
“Whoever can build the first device that can actually genuinely interact with humans, at large scale, will have huge data, extremely differentiating,” he said. “That’s why Tesla is working on a humanoid robot. They think that if they can build it under $10,000, and it’s mildly useful, tidying up your house, (and) you’ll interact with it, it’s going to be a giant data source all of a sudden.”
Alternately, he said, the model could be a drone, or a robot dog. The key is the data collection, and the automation – and the scale!
“Automated scientific engineering, design and manufacturing is also huge,” he added. “Imagine systems that could go out, mine their resources, build factories, build products, all autonomously.”
That’s the kinds of visions that he painted for us, where we really have only begun to imagine what’s possible. I think, in many people’s minds, this kind of thing sounds outlandish – but really, only because it’s never been done before! And logically, that’s not a reason that we won’t see it, sooner than later.
I thought Sertac covered a lot of ground, made our students think, and brought into focus some of the more important points on what we will be doing in the years to come.
Source link
#Moving #Bits #Moving #Beliefs #Physical
Unlock the potential of cutting-edge AI solutions with our comprehensive offerings. As a leading provider in the AI landscape, we harness the power of artificial intelligence to revolutionize industries. From machine learning and data analytics to natural language processing and computer vision, our AI solutions are designed to enhance efficiency and drive innovation. Explore the limitless possibilities of AI-driven insights and automation that propel your business forward. With a commitment to staying at the forefront of the rapidly evolving AI market, we deliver tailored solutions that meet your specific needs. Join us on the forefront of technological advancement, and let AI redefine the way you operate and succeed in a competitive landscape. Embrace the future with AI excellence, where possibilities are limitless, and competition is surpassed.