“Machines can do some things better than we can”
Heralded as “a father of the internet” Vinton G. Cerf has truly championed his discipline. Now he reveals his exciting vision for Artificial Intelligence and how to tackle the societal implications of machine learning.
Did you, as one of the “fathers of the internet”, foresee all that has become possible through the means of data collection and machine learning when you started working on what later became the Internet?
Vinton G. Cerf: Certainly, I did not foresee everything, but Bob Kahn and I did do this design with global service in mind and with expansion in all dimensions (reach, speed and applications). The design was deliberately open to new protocols and to new means of carrying data packets (e.g. optical fibre came long after the 1973 design). We also knew that the network would be a social medium, as we had seen that in the earlier ARPANET email lists. Smartphones proved to be a surprise, but we had already seen portable equipment (laptops and pads) and Alan Kay had described as early as 1968 his idea for the FLEX computer that was the proto-concept for the laptop. The system has scaled by factors of 1 to 10 million – not a bad record.
In the early days of the web there was a fear that governments could take over the internet. Now you see corporate power in the internet expanding: just a few companies such as Amazon, Facebook, Google or Apple are holding dominion over a tremendous amount of data and thereby influence over societies. Whom do we have to fear now: government or corporations?
Cerf: Governments can still take over the net – shutdowns in Togo recently and in Egypt during the Arab Spring, filtering and blocking in China, are all manifestations of that. The private sector is still highly competitive. New competitors come along all the time (think: eBay, PayPal, Yahoo!, Google, Facebook, Twitter, Snapchat, Amazon, etc). I think there is a lot to worry about in the security space (think of the recent Equifax security breach, Russian / Chinese hacking, OPM, State Department and apparent hack of NSA “tools”) and that applies to private sector and government equally.
50 years ago, the book “The New Industrial State” by Kenneth Galbraith had a similar claim. It stated that the most valuable companies at the time, due to their market value and their power over society, would supersede the government and its competences. The consequence would be a corporate state. This prediction did not come true. So, are today’s worries just the claims by Galbraith reloaded, or is the general context different this time?
Cerf: No, governments will always have the upper hand. (They have fundamental power that private enterprise does not.) This does not mean that private enterprise is powerless. Big companies with a lot of revenue and assets are capable of doing some amazing things (Think Elon Musk, SpaceX and Tesla, for example).
Also, another book, “Future Shock”, dealt with the consequences of technological progress for societies. This book by Alvin Toffler did not so much emphasise control over society but rather the impact that progress may have on a populace once it accelerates beyond the understanding of the elites and the average Joe alike. Are we in a period of time where progress has indeed reached such a pace that societies need to pause from it?
Cerf: I think if we were living in 1901, we might be having the same conversation about the telephone, electricity, automobiles, radio etc.
After the last presidential election in the United States, a data collecting company claimed to have been responsible for the outcome of the election. Is it already possible, in your opinion, to influence society that thoroughly, through microtargeting, in such a way or was this company just bragging?
Cerf: Mostly bragging, but there is clear evidence that micro-targeting works (think about advertising), so this is not a trivial matter or one to dismiss.
The developments in machine learning have raised questions about the future of work. Jobs in the hands of humans today may be taken over by algorithms and data science. This may not only affect manufacturing work, as automatisation progress has in the past; it will also affect high-skilled labour in medicine, law-making and banking. Which scenarios do you think are credible in the short and medium term?
Cerf: I think the long term has many jobs going away and many new ones created – the challenge is to re-educate people whose jobs many have been automated so they can do the new ones.
Another consequence of this development concerns the distribution of wealth. For the first time in history men may not have to work physically, manually until exhaustion, yet the question arises of what man will do, what will his or her work be, and how will he or she be compensated?
Cerf: There is a growing disparity in wealth distribution and I think that has societal risks. Whether we get to Star Trek’s 24th Century with no money is still an open question. Money has utility as a medium of exchange – it is fungible – and that’s very useful. A transition to such a state is the subject of a lot of speculative thought.
What do you personally think the future of work will be? Certainly, we will not all become painters or poets.
Cerf: I think people will continue to look for productive things to do, compensated or not. We do need basics: food, shelter, clothing, and meaningful ways to spend our time. I think that the kind of work we do will absolutely change as technology continues to evolve.
When you look back to your childhood and adolescence and recall the society you grew up in: what are the main differences when it comes to values and social norms back then and now?
Cerf: We talked to each other more back then. Privacy seemed less threatened. We were less vulnerable to social stress that seems to come from the online social media of today. On the other hand, information was harder to find – now we have it at our fingertips. We had to work a bit more to be entertained (not a bad thing). In some ways, I think we used our leisure time better – today’s world seems more broken up into smaller pieces.
To what extent do you think artificial intelligence and the developments in society that come with it, may change humanity’s perspective on itself, how may we answer the famous question by Immanuel Kant “Was ist der Mensch?”
Cerf: I think we will begin to question what it means to be intelligent – we will find that “machines” can do some things better than we can. Ultimately, I hope that artificial intelligence / machine learning will become tools that allow us to work more effectively as opposed to being competitors. I suspect we will also learn to appreciate non-human biological intelligence more – at least I hope so.
Interview by Alexander Görlach.
The interview ist part of the publication “Entering a new Era”.