AI: Futures - Oleg Akbarov

21/06/2018

Response number one to our AI: Futures discussion piece comes from Moscow based friend Oleg Akbarov who currently is Software Developer for the AI conversation application 'Replika'.

Most modern techniques of AI engineering are built around aggregating data in order to emulate an autonomous decision making process. Vast amounts of information is fed into artificial neural networks to form a 'model' which eventually will be able to make decisions later on. But where does this data come from? Who creates it? Short answer — we do. Our activity online is recorded and processed. Every little click, and time spent looking at content. Your search history and messenger logs. All is used as material to train deep artificial neural networks.

Theoretically this approach should lead to the creation of more and more advanced neural networks - ultimately, to the degree when it is impossible to distinguish artificial neural network reasoning from that performed by humans.

In general the more data you have, the better quality your AI model is. However there are only a few companies with access to massive amount of data at the moment, and you know their names. It comes as no surprise that Google, Facebook and Amazon have all made significant advances in field of AI lately. So if we are concerned about what the future of AI would look like — we should ask ourselves: can big (profit-driven) corporations be trusted with keys to probably the most advanced technology in history of humanity?

AI should be founded on genuine purpose and not as a means to grow market-share.

Currently we are complicit in this development, and if we want to avoid AI being a mere byproduct of big online business we need to find ways to hijack the process. We must become more conscious of how we share our own data and online behaviours, as well as support alternative approaches to the creations of artificial intelligence. AI should be founded on genuine purpose and not as a means to grow market-share.