Interview by Marija Butkovic @MarijaButkovic
Dr Juemin Xu is founder and CEO of Think and Decide. Combining engineering, economics and psychology, she does research in artificial intelligence and human decision-making. She is creator of AI tools for equality and empathy and her research has been covered by the Wall Street Journal, the Economist, the New Yorker and reached millions of readers.
Juemin, how did you get into the world of AI?
I did research on human behaviour. AI are idealised human cognitive models being executed by machines. The fundamental principles are based on how a "rational" person/animal/agent "should" make decisions. Then we ask the machines to apply those principles in vast speed with vast data. It's completely natural for me to move to AI from human cognitive research. We often compare human decision making to AI decision making to see who make better results. Machines are good at never being tired, never bored, 24/7, need no salary. Humans are good at being really smart, using a tiny amount of data to make amazingly accurate decisions. Researches on human are used to improve AI. For example, human heuristics is a rule of thumb, experience-based method to solve problems. It does not necessarily explore all the possibilities to produce an optimum resolution. Instead, it used a few data, produces a quick and dirty solution for an adequate result. When optimum is not required, AI heuristic model which mimicking human can achieve similar satisfying results. Human cognition can also directly be the input in AI models. For example, reCAPTCHA uses human to help recognized blurred writings from old books. Now they also ask human users to recognise pictures with cars in it to train the driverless cars.
What projects are you working on at the moment?
I'm working on a series of projects to reduce prejudice and discrimination. Everyone is naturally the "centre" of the world they know. It's difficult to understand other people's difficulties and problems. We want to create an immersive experience for a person to take another person's point of view. For example, using virtual reality and see yourself in a different colour of skin. Then the participants go to a high-end department store. The participants may see the staff treat them differently from people of other skin colour. And the participants are asked to remember this feeling and write it down later. This experience can also be used to demonstrate the different treatment received by people of different sexes, ethnic groups, religious groups etc. This can be used as a training and education tool or just for fun.
A 2nd project is also of reducing prejudice and discrimination. It is to de-polarizing political opinions. Since 2016 or even earlier, the political scene became polarized in the US, UK and other countries as well. One side feels the other side is arrogant, mad or deplorable, stupid. AI creates questions related to political tendencies, then the participants are asked to take the opposite side of the political spectrum and debate for that side. Researches have shown this method could create a middle ground of opinions. This could be made into a TV show or online debate. For this project, we are seeking business partners in the media industry.
A 3rd project is about the polarized society itself. There is evidence that the polarized society is at least partly due to the even higher efficiency created by advanced technologies including AI. Many people's jobs are replaced by machines. And the new technologies may have yet created enough new jobs to absorb the redundant workforce, nor are the new jobs suitable for the redundant workforce without substantial training. The people left behind by the technologies may find themselves without job or dignity. Maybe society should take "job" somewhat less seriously, and pay more attention to "live". Possibly universal basic income (UBI) should be available to everybody for maintaining a basic life. People could be less anxious and less hostile with a basic income. This is different from job seeker's allowance, which contains an implication of personal failure. Recent researches in Norway found that UBI did not make people work less, neither did it make people work more. It does not make people lazier. However, it makes people happier. Will UBI mend the polarized society? Will it create a less angry society? We are interested in the long term effect on people's behaviours. Will they engage more in high-risk high-return projects? Will they be more entrepreneurial? If the answer is yes, UBI could lead to a higher level of creativity in society. We could have a knowledge economy without people left behind (at least not too far). For this project, we are seeking government, public or private funding partners.
How has your career progressed since your degree? Has it been an easy industry to get into or have you had many challenges?
I founded my company Think and Decide after PhD. We are seeking funding and business partners to make the projects happen. It's never easy to be a startup. We are still bumping around to find cooperation opportunities.
How long did it take you to be where you are now?
The company was set up a year ago. We are exploring ideas in human decision making and AI years before that.
What was the biggest obstacle?
We are not sure. We want to finish the first prototype, then we will know where the obstacles are.
What are your biggest achievements to date?
Think and Decide is creating great ideas we hope will make differences. I'm looking forward to the next thing we are making.
My research in gambling was a big hit in the media. Half a million people on Twitter and millions readers on traditional media read about our research. It's rare for an academic paper to draw so much attention. We were among the top 2% papers in terms of public audience reached. The Wall Street Journal, the Economist, the New Yorker and many other media outlets reported my discovery. Gamblers unintentionally created good luck or bad luck for themselves, and in the opposite direction, i.e. if they believed they were going to be lucky, they made themselves unlucky, and vice versa.
What does the #WomenInSTEM movement mean to you? What are the challenges of being a woman in AI?
It's great to have #WomenInSTEM. There is absolutely no reason women should not be in STEM. When I grew up in China, I never heard STEM was not for women. My favourite subject at school was physics and I did engineering in my first degree. However, near graduation, I found out Chinese employers prefer men over women. Because they felt women were not suitable for business travel (physically weak, more likely to be raped) and maternity leave inconvenient for the employers. Male classmates were chosen even when I had better exam results. This was unfair. When I came to the UK, I discovered the prejudice that women were not supposed to be good at math. I was astonished. Psychological researches have shown hints of stereotype could make an actual difference in your skills. Repeatedly told girls that they are not good at STEM would certainly hinder their skill development. #WomenInSTEM is a great movement to prevent this from happening any more. I have not found too many challenges as a woman in AI. Women bring AI new perspective.
In your opinion, what will be the key trends in the AI industry in the next 5 years and where do you see it heading?
It could bring shame on oneself to predict the future. The future is notoriously resistant to predictions. Let me take the risk. I think AI decision-making assistant could be popular. For example, AI could scan through a life insurance contract and give a synopsis of the main points plus a cash flow chart. And the AI assistant can go through your past income and expense pattern and give a recommendation of the suitability of the insurance offer. In fact, this is also something I want to do, but the insurance industry is highly regulated. AI assistant can read those long and boring user agreements, terms and conditions, and report to us any unusual or unfair terms hidden inside the mile-long legal jargons.
Who are your 3 inspirational people or businesses in AI?
Judea Pearl, Herbert A. Simon, Claude Shannon
This interview was conducted by Marija Butkovic, Digital Marketing and PR strategist, founder and CEO of Women of Wearables and co-founder of Kisha Smart Umbrella. She regularly writes and speaks on topics of wearable tech, fashion tech, IoT, entrepreneurship and diversity. Visit marijabutkovic.co.uk or follow Marija on Twitter @MarijaButkovic @Women_Wearables @GetKisha.