I read Kissinger, Schmidt and Huttenlocher’s @TheAtlantic article on the future of #AI, and given the article’s reach, I think it’s worth talking about a few things. Thread.

I recommend reading the piece which I find in parts very illustrative of the debate on #AI in the policy community (I’m defining ‘policy community’ very broadly here). theatlantic.com/magazine/archi…

Of course, the authors are highly respected (though not uncontroversial) individuals, and at least two of them know *a lot more* about #AI than I do, so I mean no disrespect. Still, a few things seem worth looking at closer.

First, the framing/wording, which I am not super keen on. I don’t think that statements such as “In some ways, AI is comparable to the classical oracle of Delphi, which left to human beings the interpretation of its cryptic messages about human destiny” are terribly helpful.

Rather, they increase a sense of dread, and increase, rather than decrease, confusion.

Second - and this I found really surprising - at several instances the authors seem to suggest that they are (among) the first to think about these impacts of #AI. But there is a lot of work being done which the authors surely know, so this is a bit strange.

For instance, they note: “as a starting point - and, we hope, a springboard for wider discussion - we are engaged in framing a more detailed set of questions about the significance of AI’s development for human civilization.”
Also 👇

I was obviously most interested in the part on "AI, Grand Strategy and #Security".
The main issue I have here is that the authors do not spell out what AI-enabled military systems and weapons they imagine or refer to, but rather talk about #AI in general.

But, as @mchorowitz has laid out very well, AI is an enabler, not one single technology. The impact on warfare of AI-enabled autonomous drone swarms might be very different than that of AI-enabled #cyber weapons, or AI-enabled surveillance.
tnsr.org/2018/05/artifi…

Thus, when they talk about strategic implications of AI in warfare, I kept wondering which AI-enabled systems or weapons they were thinking about. The answer to these 👇 questions for example, really depend on what you are looking at.

Here again, it seems that they way they chose to write this, adds to the confusion and sense of danger for the reader. Also, they don’t seem to attempt to differentiate between short-term and long-term implications which might also be helpful.

To be clear: I find it very hard to wrap my head around these issues, which I am trying to do every day. And criticising others' work is much easier that doing it better yourself. But maybe thinking through the points above can help us all be more precise.

So, definitely check out the article – and I’d be very interested in other people’s takes on this. theatlantic.com/magazine/archi…

And if you’re not already familiar with it, also read
tnsr.org/2018/05/artifi…

Finally, I recommend @paul_scharre's* great thread on the same article, in which he focuses, among other things, on the definition of intelligence. Definitely worth reading.

https://twitter.com/paul_scharre/status/1150902676743479297

*who is one of the people who are working tirelessly to make sense of all of this