The big problem with big data? Without theory, it's just garbage

Uta Frith doesn’t want to meet Donald Trump. “There would be no point in my saying anything to him,” she says. “Mostly, when scientists give advice to politicians, politicians listen only to the things they want to hear.” Frith, a developmental psychologist who works at University College London, should know. Not only has she been a pioneer in the study of dyslexia and autism — in the 1960s, she was one of the first researchers in the UK to study Asperger's Syndrome — but she has also been working to advance the interests of women in science for decades.

As part of our Scientists Meet the Media series, Frith spoke to WIRED about mind over matter, big data and her fear that AI will lead to loss of meaningful jobs.

Being anti-science can only too easily be combined with embracing the benefits of science. The rapid advances are perceived as part of technology, and technology is used because it works. If you use a gadget or a pill and it works, you don’t concern yourself with the question, “How does it work?”. This question is left to the scientists to answer. But their answers are often dismissed as just theory. There is a suspicion that scientists themselves don’t know the answer, and merely guess (stating probabilities rather than certainties). They are also known to argue among themselves, which only reinforces the idea.

Perhaps this state of affairs is also due to the fact that scientific explanations are very hard to understand. We are good at avoiding difficult tasks and at justifying this. So, rather than admitting to not getting it, it is comforting, at least temporarily, to stand by one’s own irrational opinion and to dismiss scientists’ theories as abstruse or self-serving.

What I’d dearly love to see before I die is a well worked example going all the way from gene to neuron to brain system to explain a phenomenon we can call ‘mind over matter’. For example, how an idea verbally expressed by one person can result in a complex set of thoughts and consequent behaviour in another person. If we knew how this works perhaps we could accelerate learning from others.

I fear the loss of occupations without replacement of meaningful activity. This is obviously an unintended consequence of AI development, and something that we should try and prevent. If large sections of the population are left unprepared for being without a meaningful job, there could be civil unrest, even if basic material needs were supplied.

Data is often full of errors and the bigger the set the more errors there are likely to be. I am sceptical of big data if there is no prior hypothesis, and I fear that the results will be meaningless if big data are analysed in purely bottom-up fashion. “Garbage in – garbage out”. Interpretations can be distorted by random events. The idea that big data will enable more control of behaviour may be a lot of hype.

The world would be more just and more fair. This is worth it for its own sake. But there is also a utilitarian argument. Increased diversity in business is economically advantageous, and increased diversity in science teams will help us create better models of the world. I sometimes wonder why the moral argument is not more strongly supported, because we all benefit from a fair and just society.

Old age and its physical and mental hazards. Problems associated with ageing are becoming more obvious as life expectancy is extended. They include psychiatric and neuropsychological problems, as the brain is ageing just as much as the body. This is possibly under-recognised at present.

Measures need to be developed that properly assess the quality of life of an ageing person, in terms of physical and mental well being. Just to keep someone alive without regard of their wellbeing seems to me irresponsible. Death is part of life, and science needs to help us understand what this really means for individuals, and how we best cope with this fact.

Frith was among the guests at the recent Scientists Meet the Media reception at the Royal Society in London, which is organised with the Science Museum. The event was sponsored by Johnson & Johnson Innovation and supported by the Association of British Science Writers and WIRED.