AI is not human, and never will be

In 1970 an American Mathematician by the name of Marvin Minsky was interviewed by journalist Brad Darrach for Life magazine. In this interview Minsky stated “In from three to eight years we will have a machine with the general intelligence of an average human being.” further going on to explain ‘If we’re lucky, they might decide to keep us as pets.’ Since the 1950’s AI mathematicians, scientists and researchers, many of them rationalists, have been trying to achieve Artificial General Intelligence. Artificial intelligence that thinks, feels and behaves like a human being. Pop-culture, Sci-fi and Hollywood has for decades pushed an image of AI that far exceeds the intellectual abilities of a human brain. Portraying scenes of humanity becoming slaves to data fuelled semiconductors. 55 years on from Minsky’s interview, We’re now in a world where this is reasonably possible and doesn’t sound as absurd as it would’ve in previous decades.

There is a common pattern amongst these AI researchers, scientists and entrepreneurs, they regard humans and AI as the same thing, the same category. To these rationalists humans and artificial intelligence are both systems that input data, make some calculations and give an output. You can see this in the way they talk about AI and the future they envision. But, AI will never be human. Never. Because it is man-made it is in the shadow of us. AI is a mirror reflecting back at us our aggregate thoughts, actions and knowledge. It is a very accurate probability indicator. AI robots may use sensors to predict when they’ve been touched but simulating embodiment is not the same as embodiment itself. It does not have sensory neurons or an amygdala and the causal chain to make neurochemical changes that have evolved over millions of years. Nothing that is made can be in the same category as the maker.

My point is not to talk down the development of AI but to say that we need to re-align our objectives on it. The goal of trying to make AI like a human is futile. We should instead be leveraging it to help us as a species solve our biggest problems like cancer, climate change and fusion energy. Let’s move away from the Hollywood sci-fi narrative and push for a more positive and progressive one where AI is used for one reason, to improve people's lives!

There are currently many labs using AI to identify cancer biomarkers, modelling climate systems at a resolution not possible before. This is AI being used for good. AI used as a tool. We don’t need AI to be human. We need it to aid us in doing what we couldn’t before. The obsession with AGI is a distraction. A distraction that is focused on automating and replacing humans.

Now, instead of the central question being  ‘How do we make AI more human?’ I think it should be ‘How do we use AI to augment humans?’ If AI is a tool, how can we design systems and infrastructure for us to use it in the most effective way possible to be able to solve problems in ways we’ve never thought possible or thought of at all. What I'm expressing here is Human-Centered AI (HCAI) thinking. AI products, tools and systems that are centered around enhancing, augmenting and extending human capabilities in a safe, reliable and trustworthy way.