29.04.2019Blog

AI – to bias or not to bias?

Solita The Impact of AI report Q&A Sonal Makhija

Could AI make the world even more unequal? Should we simply accept the possibility that AI models will replicate pre-existing human biases? What choices can we make today to ensure that AI does not worsen prejudice and bias? Transparent debate is needed more than. We sat down with Sonal Makhija, a lawyer and anthropologist, and contributor to our recent report The Impact of AI, and examined some of the most crucial issues facing our collective future.

Do you think it is acceptable for AI systems to be biased and who is responsible for preventing this from happening?

I do not think it is acceptable that AI would be biased. I do not think it would acceptable if you put a human in that place. Of course, we recognise that there is no pure objectivity and there will be biases. But the conversation and the expectation of using AI is that it is going to be accurate and objective. It is expected that AI eliminates our errors and fallacies and biases.

If we build a machine that instead of eliminating our biases, actually replicates and mimics them – and does this even more accurately – then it is highly problematic.

We would not want that kind of intelligence. Or at least, I do not consider that intelligence.

As to who is responsible, I think companies do need to have a conversation around this.

Before building and bringing new solutions into the market, there has to be a risk assessment.

Civil society, social scientists and policy experts need to be involved.

I think we cannot let big private entities completely hijack or define what is good for us. I think if anything, with the manner in which things have gone ahead with, for example, social media – that is something that we can learn from. Organisations are still struggling with and have not been able to grapple with what is happening with social media. From fake news to privacy issues and safety – it is extremely important to be a part of this conversation right from the beginning. Technology cannot fix every problem, we need human interventions.

How do you see AI’s potential for humankind in the future?

The reason why we are having this discussion is because we do see there is enormous potential for us as human beings. Such as, in terms of healthcare, hopefully creating a more equal society. Things like self-driving cars can mean greater mobility for the disabled. And it can mean cheaper travel.

So, of course, there is great potential, but much of it depends on what direction it takes and that depends on us.

Sonal Makhija is an anthropologist and lawyer based in Helsinki. She is intrigued by the cultural and social effects that AI will have on humans.

Impact of AI