Today I was having a conversation on Twitter on which future would be more frightening, one where politicians think they’re smarter than big data – which is more like the present – or one where we put in power an all powerful AI to take these decisions for us, without it (he/she?) being held accountable by humans.
I jokingly replied that – after seeing the Matrix and Terminator series – I would prefer the greedy but human politician. And later, while in that place where…you know, most of humankind’s thinking takes place :)), I ended up asking myself: would I? Yes I would! And here’s why…
A bit of context
I’ll start with the definition for big data – nicked from Wikipedia:
Big data is a broad term for data sets so large or complex that traditional data processing applications are inadequate. Challenges include analysis, capture, data curation, search, sharing, storage, transfer, visualization, querying and information privacy. The term often refers simply to the use of predictive analytics or certain other advanced methods to extract value from data, and seldom to a particular size of data set.
The most traditional way of interacting with stored data is by looking at it. Reading it and drawing conclusions for oneself. This no longer applies, as every day we generate huge amounts of data, in the range of exabytes. That’s more information that our species generated – if we are to consider every story, every legend, every gesture ever made by every member of every civilisation – since the beginning of time until the invention of the internet. Even the most skilled speed-readers cannot hope to finish reading that in the lifespan of the…Universe. So we use machine learning to process the data and act under the assumption that the end result is the correct one, often excluding the possibility of a bug in the application or that the developers acted in bad faith.
If we have 2 big data systems that yield conflicting results, since we can’t really check the data, we will just pick the one that validates our existing beliefs.
The stupid & greedy politician
We’re getting frustrated when politicians deny data that we perceive to be undeniable, often at the pressure – financial or otherwise – of big interest groups. Global warming, wars for resources, embargos and laws designed to help a small influential group profit and so on. We see this as wrong and anti-democratic. The little guy suffers (which is not true, the global standard of living has been constantly on the rise and we’re living in the best period in recorded history – that’s big data for you 😉 ).
And while this is bad, we know from experience that dictatorships are worse. But what if the dictator is an AI that’s unaccountable to humans?
I don’t think the AI would be fit to govern. Because even if we would overcome the technical challenges and actually build a super intelligent AI capable of dealing with the complexity of running the planet, I don’t think it would be capable to relate to humans well enough to run our society. Such an agent will be completely cut off from humanity, from the needs of the many. Even the brightest humans will seem like ants to it. The AI would be even more disconnected from the needs of the little guy than the politician. And from the needs of humanity as a whole. There’s a whole lot that can go wrong. Here’s a quote from one of the leading experts in AI research:
The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else.
And in the end, the politician wants to be re-elected…