THE QUEST FOR ETHICAL ARTIFICIAL INTELLIGENCE: A CONVERSATION WITH TIMNIT GEBRU

Harvard Radcliffe Institute hosted a zoom webinar ‘The Quest for Ethical AI’ on Mar 8, 2022 at 4 PM ET. In this webinar, Timinit Gebru, founder and executive director of Distributed Artificial Intelligence Institute, discussed the reasons why she founded DAIR and what it should accomplish.

DAIR institute is composed of an interdisciplinary team that is free from the influence of all “big” technology. This company is composed of four core values as discussed by Timinit:

1. Community, not exploitation

   • I found this value as the most important and attractive of them all. I believe researching to bring diversity and inclusion in AI and all tech fields is important.

2. Healthy, thriving researchers

   • Supporting researchers to stay healthy is also important in all kinds of areas.

3. Comprehensive, principled process

   • This value helps conduct a quality finding which is the ultimate goal of any research.

4. Proactive and pragmatic research

   • This is the second most important value in my perspective as it prevents technology with potentially harmful effects.

The purpose of this institute is interesting as it is fighting against the ethical issue with AI technology. Timinit explained the reason why she uses the term ‘Ethical AI’ instead of ‘AI for social good’ as “Because I want all AI to be good.” I like the explanation even though it sounds unrealistic as she also said that the “new AI bloom is very much tied with how much you can make” from the technology therefore sometimes the harm it causes is overlooked. One solution she proposed for this issue is to advocate for research publication venues so that we create an incentive for researchers to work more on this topic. I agree with this solution as it will potentially bring more diverse solutions to the ethical issues of AI. 

Timinit said the vision she has for DAIR institute in the next five years is two:

1. To proactively work for AI research to benefit the community and focus on having healthy and thriving researchers.

2. Do work that clearly shows the harms of AI technology

The vision is no different from the core values of the institute. Though I expect it to change as they gradually grow the institute. 

One of the best tips Timinit gave for students doing theoretical research is to consider three things:

• What is the question I am asking?

• How am I trying to answer it with a theoretical sense and what can we do to experiment to see where we are?

• To pull back and question the norms on our field

  • I believe this is the starting point for almost all researchers because it is when we thoroughly observe that we start to see what is wrong and move to find solutions.

As of right now (3/8/2022), DAIR institute is not hiring any kinds of positions but Timinit said one can follow up on Twitter and by signing up for the newsletter as well.