How smart can our machines make us? In his talk at TED 2017, Tom introduces the idea of Humanistic AI that augments and collaborates with us instead of competing with or replacing us. He shares his vision for a future where AI helps us achieve superhuman performance in perception, creativity and cognitive function — from turbocharging our design skills to helping us remember everything we’ve ever read. The idea of an AI-powered personal memory also extends to relationships, with the machine helping us reflect on our interactions with people over time. The upside? “Every time a machine gets smarter, we all get smarter.”
Here is the full transcript. Please visit the TED site for the transcript in twenty languages.
I’m here to offer you a new way to think about my field, artificial intelligence. I think the purpose of AI is to empower humans with machine intelligence. And as machines get smarter, we get smarter. I call this “humanistic AI” — artificial intelligence designed to meet human needs by collaborating and augmenting people. Now, today I’m happy to see that the idea of an intelligent assistant is mainstream. It’s the well-accepted metaphor for the interface between humans and AI. And the one I helped create is called Siri.
You know Siri. Siri is the thing that knows your intent and helps you do it for you, helps you get things done. But what you might not know is that we designed Siri as humanistic AI, to augment people with a conversational interface that made it possible for them to use mobile computing, regardless of who they were and their abilities.
Now for most of us, the impact of this technology is to make things a little bit easier to use. But for my friend Daniel, the impact of the AI in these systems is a life changer. You see, Daniel is a really social guy, and he’s blind and quadriplegic, which makes it hard to use those devices that we all take for granted. The last time I was at his house, his brother said, “Hang on a second, Daniel’s not ready. He’s on the phone with a woman he met online.” I’m like, “That’s cool, how’d he do it?” Well, Daniel uses Siri to manage his own social life — his email, text and phone — without depending on his caregivers. This is kind of interesting, right? The irony here is great. Here’s the man whose relationship with AI helps him have relationships with genuine human beings. And this is humanistic AI.
Another example with life-changing consequences is diagnosing cancer. When a doctor suspects cancer, they take a sample and send it to a pathologist, who looks at it under a microscope. Now, pathologists look at hundreds of slides and millions of cells every day. So to support this task, some researchers made an AI classifier. Now, the classifier says, “Is this cancer or is this not cancer?” looking at the pictures. The classifier was pretty good, but not as good as the person, who got it right most of the time.
But when they combine the ability of the machine and the human together, accuracy went to 99.5 percent. Adding that AI to a partnership eliminated 85 percent of the errors that the human pathologist would have made working alone. That’s a lot of cancer that would have otherwise gone untreated. Now, for the curious, it turns out that the human was better at rejecting false positives, and the machine was better at recognizing those hard-to-spot cases. But the lesson here isn’t about which agent is better at this image-classification task. Those things are changing every day. The lesson here is that by combining the abilities of the human and machine, it created a partnership that had superhuman performance. And that is humanistic AI.
Now let’s look at another example with turbocharging performance. This is design. Now, let’s say you’re an engineer. You want to design a new frame for a drone. You get out your favorite software tools, CAD tools, and you enter the form and the materials, and then you analyze performance. That gives you one design. If you give those same tools to an AI, it can generate thousands of designs.
This video by Autodesk is amazing. This is real stuff. So this transforms how we do design. The human engineer now says what the design should achieve, and the machine says, “Here’s the possibilities.” Now in her job, the engineer’s job is to pick the one that best meets the goals of the design, which she knows as a human better than anyone else, using human judgment and expertise. In this case, the winning form looks kind of like something nature would have designed, minus a few million years of evolution and all that unnecessary fur.
Now let’s see where this idea of humanistic AI might lead us if we follow it into the speculative beyond. What’s a kind of augmentation that we would all like to have? Well, how about cognitive enhancement? Instead of asking, “How smart can we make our machines?” let’s ask “How smart can our machines make us?” I mean, take memory for example. Memory is the foundation of human intelligence. But human memory is famously flawed. We’re great at telling stories, but not getting the details right. And our memories — they decay over time. I mean, like, where did the ’60s go, and can I go there, too?
But what if you could have a memory that was as good as computer memory, and was about your life? What if you could remember every person you ever met, how to pronounce their name, their family details, their favorite sports, the last conversation you had with them? If you had this memory all your life, you could have the AI look at all the interactions you had with people over time and help you reflect on the long arc of your relationships. What if you could have the AI read everything you’ve ever read and listen to every song you’ve ever heard? From the tiniest clue, it could help you retrieve anything you’ve ever seen or heard before. Imagine what that would do for the ability to make new connections and form new ideas.
And what about our bodies? What if we could remember the consequences of every food we eat, every pill we take, every all-nighter we pull? We could do our own science on our own data about what makes us feel good and stay healthy. And imagine how this could revolutionize the way we manage allergies and chronic disease.
I believe that AI will make personal memory enhancement a reality. I can’t say when or what form factors are involved, but I think it’s inevitable, because the very things that make AI successful today — the availability of comprehensive data and the ability for machines to make sense of that data — can be applied to the data of our lives. And those data are here today, available for all of us, because we lead digitally mediated lives, in mobile and online.
In my view, a personal memory is a private memory. We get to choose what is and is not recalled and retained. It’s absolutely essential that this be kept very secure.
Now for most of us, the impact of augmented personal memory will be a more improved mental gain, maybe, hopefully, a bit more social grace. But for the millions who suffer from Alzheimer’s and dementia, the difference that augmented memory could make is a difference between a life of isolation and a life of dignity and connection.
We are in the middle of a renaissance in artificial intelligence right now. I mean, in just the past few years, we’re beginning to see solutions to AI problems that we have struggled with literally for decades: speech understanding, text understanding, image understanding. We have a choice in how we use this powerful technology. We can choose to use AI to automate and compete with us, or we can use AI to augment and collaborate with us, to overcome our cognitive limitations and to help us do what we want to do, only better. And as we discover new ways to give machines intelligence, we can distribute that intelligence to all of the AI assistants in the world, and therefore to every person, regardless of circumstance. And that is why, every time a machine gets smarter, we get smarter.
That is an AI worth spreading.