A conversation with George John, CEO of Rocket Fuel, who gained rocket scientist credentials at NASA and puts his know-how to use in the world of advertising and branding.
As DLD is turning ten in 2014, what were some of the most important developments in your field in the last decade? And in your view - what's the most exciting development in AI ahead of us?
I think big data is the most important milestone. The granularity with which we can remember and analyse data now for decision making is immense. Rocket Fuel would not have been possible ten years ago. With regards to AI I can’t think of a ten year horizon, but it will go as far as anyone has been able to imagine. My bet is within our life time we will see, maybe not physical robots that resemble humans, but we’re likely to see special purpose AI that will be functional in a certain domain, which can be as varied as game shows, medical or legal AI, where a huge amount of information needs to be processed but with the ability to reason. So fields where it’s still difficult to imagine, AI will be used in the near enough future.
Can you share some of your insights on how AI may reach limits in terms of creativity? AI may allow us to build machines and robots that are capable of incredible things, but can they ever reach human capacity of creative thinking?
There’s a story in a book called “Automate This” by Chris Steiner and it tells of someone who programed an AI system to compose music. When the music was played to an audience they all thought it was wonderful, until they were told that a computer had made it which made them mad. So, my guess is if you can train an AI system to approximate the appeal of a work of art to humans, then the system could produce, first randomly and eventually more and more productively, artefacts that would be appealing to humans as art. Of course one could get really philosophical about this and ask what makes us human and is creative thought one of those features? But so far the history of AI is that there is a level of human conceipt. So we think of a task that only we can do, and in the history of AI sure enough someone has programed a computer to do just that thing. But then there’s the “moving frontier problem” in AI, which is that by the time we have programed a computer to do a very human task it’s not really considered AI anymore. Like navigation systems in cars nowadays are just a normal feature of a modern car and not some amazig feat of artifical intelligence.
Putting big data in the context of advertising - how powerful is it? And how aware do you think the average consumer is of how his or her data is used?
Advertisers have been rational for long time. You know they run a campaign and assess how it goes and try to do something even smarter the next time. The advance with big data is that you don’t only see how a whole campaign went but you can assess how a single exposure works. So I think the value is to be able to observe and learn at a much finer level. With regards to consumers I think it’s interesting, because I don’t encounter many who have the right concept of anonymous tracking for instance. There’s a false idea among some that they are being tracked and there’s someone with a whole dossier on them in some far away office. But my hope is that the industry is changing and moving towards initiatives that give the consumer more control in how they can be tracked, so that there’s more of an understanding for the algorithmically curated web experience. To give you an example: we rarely get emails regarding privacy at Rocket Fuel. But once in a while we do get an email which is really angry about being anonymously tracked, and people give us their name and email address and ask to be deleted from the system. Ironically we could not have known their name or email address up to this point. So I think there’s a gap in how consumers use the internet and how they understand it.
Is Big Data, and the understanding companies can galvanise from it, applicable to any kind of brand or do some brands have to rely on different ways to understand their customer base?
Well I think big data always helps but it’s a question of how much granularity makes sense. You can definitely create a lot of extra value from the context in which the data is collected. Even if you stay away from specific information about the consumer, information that remains anonymous, you still get a better understanding of your customer base.
What are the most important and maybe unexpected skills you bring from your time at NASA as a rocket scientist to your job as CEO at Rocket Fuel?
At NASA I was in the group called “artifical intelligence lab”, and the main topics we dealt with were related to either autonomous space crafts or other kinds of AI related to augmenting human work. When working on autonomous space craft you have a certain style of thinking, because you don’t think of a human pilot who you can advise how to use the space craft. You think of a fully autonomous auto pilot that makes all decisions by itself, which leads you to take a much broader range of possibilities into account. So I think that was useful for Rocket Fuel and our advertising systems, because these systems have up to forty billion opportunities a day to reach out to consumers, which is well beyond the capacity of any one human. In my view, doing that right demanded the same kind of autonomous space craft way of thinking.
George John will speak at the upcoming DLD14 conference, taking place in Munich January 19 - 21, 2014. Apply for a ticket to this exclusive conference, tune in on the beat of our community on the DLDpulse and find regular updates on the DLD14 programme and speakers here.