The story of AI, as instructed by the individuals who invented it


Welcome to I Was There When, a brand new oral historical past venture from the In Machines We Belief podcast. It options tales of how breakthroughs in synthetic intelligence and computing occurred, as instructed by the individuals who witnessed them. On this first episode, we meet Joseph Atick— who helped create the primary commercially viable face recognition system.


This episode was produced by Jennifer Sturdy, Anthony Inexperienced and Emma Cillekens with assist from Lindsay Muscato. It’s edited by Michael Reilly and Mat Honan. It’s blended by Garret Lang, with sound design and music by Jacob Gorski.

Full transcript:


Jennifer: I’m Jennifer Sturdy, host of In Machines We Belief

I wish to inform you about one thing we’ve been engaged on for a short time behind the scenes right here. 

It’s known as I Was There When.

It’s an oral historical past venture that includes the tales of how breakthroughs in synthetic intelligence and computing occurred… as instructed by the individuals who witnessed them.

Joseph Atick: And as I entered the room, it noticed my face, extracted it from the background and it pronounced: “I see Joseph” and that was the second the place the hair on the again… I felt like one thing had occurred. We have been a witness. 

Jennifer: We’re kicking issues off with a person who helped create the primary facial recognition system that was commercially viable… again within the ‘90s…


I’m Joseph Atick. Right this moment, I am the chief chairman of ID for Africa, a humanitarian group that focuses on giving folks in Africa a digital id to allow them to entry companies and train their rights. However I’ve not all the time been within the humanitarian area. After I acquired my PhD in arithmetic, along with my collaborators made some basic breakthroughs, which led to the primary commercially viable face recognition. That is why folks check with me as a founding father of face recognition and the biometric trade. The algorithm for the way a human mind would acknowledge acquainted faces grew to become clear whereas we have been doing analysis, mathematical analysis, whereas I used to be on the Institute for Superior Research in Princeton. However it was removed from having an concept of how you’ll implement such a factor. 

It was an extended interval of months of programming and failure and programming and failure. And one evening, early morning, really, we had simply finalized a model of the algorithm. We submitted the supply code for compilation as a way to get a run code. And we stepped out, I stepped out to go to the washroom. After which once I stepped again into the room and the supply code had been compiled by the machine and had returned. And often after you compile it runs it robotically, and as I entered the room, it noticed a human shifting into the room and it noticed my face, extracted it from the background and it pronounced: “I see Joseph.” and that was the second the place the hair on the again—I felt like one thing had occurred. We have been a witness. And I began to name on the opposite individuals who have been nonetheless within the lab and every one among them they’d come into the room.

And it might say, “I see Norman. I’d see Paul, I’d see Joseph.” And we might form of take turns operating across the room simply to see what number of it may well spot within the room. It was, it was a second of fact the place I’d say a number of years of labor lastly led to a breakthrough, regardless that theoretically, there wasn’t any further breakthrough required. Simply the truth that we discovered how you can implement it and eventually noticed that functionality in motion was very, very rewarding and satisfying. We had developed a staff which is extra of a growth staff, not a analysis staff, which was targeted on placing all of these capabilities right into a PC platform. And that was the delivery, actually the delivery of business face recognition, I’d put it, on 1994. 

My concern began in a short time. I noticed a future the place there was no place to cover with the proliferation of cameras in all places and the commoditization of computer systems and the processing skills of computer systems turning into higher and higher. And so in 1998, I lobbied the trade and I stated, we have to put collectively rules for accountable use. And I felt good for some time, as a result of I felt we’ve gotten it proper. I felt we have put in place a accountable use code to be adopted by no matter is the implementation. Nonetheless, that code didn’t reside the take a look at of time. And the rationale behind it’s we didn’t anticipate the emergence of social media. Mainly, on the time after we established the code in 1998, we stated crucial component in a face recognition system was the tagged database of recognized folks. We stated, if I am not within the database, the system will probably be blind.

And it was tough to construct the database. At most we might construct thousand 10,000, 15,000, 20,000 as a result of every picture needed to be scanned and needed to be entered by hand—the world that we reside in as we speak, we are actually in a regime the place we’ve allowed the beast out of the bag by feeding it billions of faces and serving to it by tagging ourselves. Um, we are actually in a world the place any hope of controlling and requiring all people to be accountable of their use of face recognition is tough. And on the identical time, there isn’t a scarcity of recognized faces on the web as a result of you may simply scrape, as has occurred just lately by some corporations. And so I started to panic in 2011, and I wrote an op-ed article saying it’s time to press the panic button as a result of the world is heading in a path the place face recognition goes to be omnipresent and faces are going to be in all places obtainable in databases.

And on the time folks stated I used to be an alarmist, however as we speak they’re realizing that it is precisely what’s occurring as we speak. And so the place can we go from right here? I have been lobbying for laws. I have been lobbying for authorized frameworks that make it a legal responsibility so that you can use someone’s face with out their consent. And so it is now not a technological situation. We can’t include this highly effective know-how via technological means. There needs to be some form of authorized frameworks. We can’t permit the know-how to go an excessive amount of forward of us. Forward of our values, forward of what we predict is suitable. 

The problem of consent continues to be one of the vital tough and difficult issues when it offers with know-how, simply giving someone discover doesn’t imply that it is sufficient. To me consent needs to be knowledgeable. They’ve to grasp the results of what it means. And never simply to say, nicely, we put a enroll and this was sufficient. We instructed folks, and if they didn’t wish to, they may have gone anyplace.

And I additionally discover that there’s, it’s so simple to get seduced by flashy technological options that may give us a short-term benefit in our lives. After which down the road, we acknowledge that we have given up one thing that was too valuable. And by that cut-off date, we’ve desensitized the inhabitants and we get to a degree the place we can’t pull again. That is what I am anxious about. I am anxious about the truth that face recognition via the work of Fb and Apple and others. I am not saying all of it’s illegitimate. Plenty of it’s official.

We have arrived at a degree the place most people might have change into blasé and should change into desensitized as a result of they see it in all places. And perhaps in 20 years, you step out of your own home. You’ll now not have the expectation that you simply would not be not. It won’t be acknowledged by dozens of individuals you cross alongside the way in which. I believe at that cut-off date that the general public will probably be very alarmed as a result of the media will begin reporting on circumstances the place folks have been stalked. Folks have been focused, folks have been even chosen primarily based on their web value on the street and kidnapped. I believe that is lots of accountability on our fingers. 

And so I believe the query of consent will proceed to hang-out the trade. And till that query goes to be a end result, perhaps it will not be resolved. I believe we have to set up limitations on what might be completed with this know-how.  

My profession additionally has taught me that being forward an excessive amount of isn’t a superb factor as a result of face recognition, as we all know it as we speak, was really invented in 1994. However most individuals suppose that it was invented by Fb and the machine studying algorithms, which are actually proliferating everywhere in the world. I principally, sooner or later in time, I needed to step down as being a public CEO as a result of I used to be curbing the usage of know-how that my firm was going to be selling as a result of the concern of damaging penalties to humanity. So I really feel scientists have to have the braveness to venture into the long run and see the results of their work. I am not saying they need to cease making breakthroughs. No, you need to go full pressure, make extra breakthroughs, however we also needs to be sincere with ourselves and principally alert the world and the policymakers that this breakthrough has pluses and has minuses. And due to this fact, in utilizing this know-how, we want some form of steerage and frameworks to ensure it is channeled for a optimistic utility and never damaging.

Jennifer: I Was There When… is an oral historical past venture that includes the tales of people that have witnessed or created breakthroughs in synthetic intelligence and computing. 

Do you’ve got a narrative to inform? Know somebody who does? Drop us an e-mail at



Jennifer: This episode was taped in New York Metropolis in December of 2020 and produced by me with assist from Anthony Inexperienced and Emma Cillekens. We’re edited by Michael Reilly and Mat Honan. Our combine engineer is Garret Lang… with sound design and music by Jacob Gorski. 

Thanks for listening, I’m Jennifer Sturdy. 



Please enter your comment!
Please enter your name here