stem cells in their environment

The Robots are coming (... or are they?)

06 June 2015
By Davide Danovi

'We are lived by powers we pretend to understand' - WH Auden

Although you are never sure here, the summer seems to have finally decided to hit London this weekend as we are busy with data and preparing our high content event.

Our field like many others is a testimony of the synergistic 'force beyond calculation' derived from the marriage between humans and computers as in the famous quote misattributed to Albert Einstein. It is indeed a joy that by means of principal component analysis or other unsupervised methods, an automated system can suggest something unique is happening in a particular sample, conditions or well out of the hundreds or thousands or more analysed. Machine learning can then be applied to bin objects into different categories ultimately purely starting from pixel intensities. It works for cells and it may work for people on close circuit television in a shopping mall over months lasting time frame videos.

Whether unsupervised or supervised (or semi-supervised) is your preferred choice in life, there is a lot to learn from computers, about computers and about how they can learn from us. With this spirit I visited the keynote lecture of the brilliant conference 'Machine Intelligence 2015' organised by the centre alumnus Nathan Benaich now working in Playfair Capital.

Mustafa Suleyman from Google Deepmind gave a great talk about 'Applied artificial general intelligence'. He showed how computers now play Atari games generally better than humans (and they even manage to get a Nature cover for this).

He also gave a sense of how artificial intelligence can be applied to image recognition and how the tools are implemented in searches. He mentioned low hanging fruits in health where all sorts of data can be used for diagnosis and prognosis. Interestingly he discussed the big challenges that humanity is facing such as lack of clean water, massive inequality and financial crises and climate change. He favoured the idea that artificial intelligence could help mankind.

In a nutshell, he was overall quite dismissive about singularity theory and its existential risks. I remember chatting with Sabine Hauert who knows way more than me about these topics and she seemed to share a similar frustration about the constant apocalyptic focus taking interest away from the great potential achievements. Maybe it's similar to when we have to reassure people that stem cells are not there in our incubators to be used for evil purposes yet I can't help but remaining a bit uncertain of the overall picture. Before machines are ready to take over by themselves I personally find it scary that a mind-boggling amount of information collected around and about each of us could be used (and partly is being used already) to stratify our preferences and target us. I am just genuinely not reassured it is always for our own good.

What if there was just a bit more about being human? The old debate between reductionist and holistic visions of biology and of the whole world echoes in two recent movies. A warm dialogue with humans is at the core of the operating systems (in Her) or the robots (in Ex Machina) gaining identity and consciousness. I do wish that Skynet and the Matrix remain bad nightmares and we land on our feet with this whole thing. I hope that we treasure the potential for health and wealth and feed the new challenges with robust Ethics. And that we even learn through the whole process of teaching to entities we built from scratch.

'What I cannot create, I do not understand' - Richard Feynman

comments powered by Disqus