In early November, AITAB (AI Theology Advisory Board) met for another fruitful conversation. As in our previous meeting, the dynamic interaction between our illustrious members took the dialogue to places I did not anticipate. In this first part, we set up the dialogue by framing the key issues in AI for good. We then move to a brief report on how this is playing out in East Asia. In a conversation often dominated by Silicon Valley perspectives, it is refreshing to take a glimpse at less-known stories of how AI technologies are reshaping our world.
Defining AI for Good
Elias: Let me say a few words to start the discussion. This is different from our other discussions where we focused on ethics. In those instances, we were reflecting on what was “wrong” with technology and AI. Today, I wanted to flip this script and focus more on the positive, what I call “AI for good”. Good theology starts with imagination. Today is going to be more of an exercise of imagination to notice what’s happening but doesn’t necessarily make the news.
More specifically, there are three main areas where I see global AI for good starting to take shape. The first is the democratization of AI skills – spreading technological knowledge to underrepresented communities. This is a very important area, since as we discussed before, people make AI in their own image. If we don’t widen the group, we will have the same type of AI. A great example is Data Science Nigeria. Just yesterday I was in one of their bootcamps as a speaker. It’s very encouraging to see young men and women from Nigeria and other African countries getting involved in data science. It started as a vision of two data scientists that want to train 10 million data scientists in Nigeria for the next 10 years. It’s a bold goal, and I sure pray they achieve it.
The other topic is about Green AI or Sustainable AI. How AI can help us become more sustainable. One example is using computer vision to identify illegal fires in the Amazon – using AI to affect change with an eye on sustainability. And the last one is AI justice. The same way AI is creating bias, it’s using AI tools to identify and call out this bias. That is the work of some organizations like Algorithmic Justice League led by Joy Buolamwini. That’s also an area that is growing. These three areas cover the main themes of global AI for good.
Re-Framing Technology
Let me frame them within a biblical context. In technology when we usually mean big tech that comes from Silicon Valley. As an alternative, I want to introduce this different concept, which is mustard seed technology. In the gospels, Jesus talks of the kingdom of God being like a mustard seed. Though it’s one of the smallest seeds it becomes a big tree where birds can come and rest in their shade.
I love this idea about this grassroots technology, either being developed or being deployed to provide for others. Just think of farmers in Kenya using their phones to make payments and doing other things they haven’t been able to do before. Those are the stories I wanted to think about today. I wanted to start thinking geographically. How does global AI for good look like in different places of the world?
AI for Good in East Asia
Levi : Here in East Asia, the turning point came in 2016 when DeepMind AlphaGo (Google supercomputer) beat Lee Se Dol in a game of Go. It created a very interesting push in South Korea and China to rapidly advance and develop AI infrastructures. I’m involved with a group on AI social and ethical concerns focused on Asia. The group has nine different scholars from 6 different Asian countries. One of the things we are going to discuss soon is a report from MIT from interviewing several Asian business owners about direction. This report is 2 years old, but it’s interesting to see how small focused the state of China was then. Now they are one of the world leaders in AI development.
There is a major push in this part of the world. Asia across the board was late to the industrial game, except for Japan. As many countries like South Korea, China have massively industrialized in the last decades, they see AI as a way to push into the future. This opens a lot of questions. Like the ones about democratization and justice that need to be addressed. But one of the interesting things is that Asian countries are interested in pushing towards AI regulation compared to the USA or other European countries. There is also this recognition of wanting to be the best in advanced technology but also the best in “getting it right”.
Where that’s going to go it’s hard to say. We know for that in China, the government directs most of AI development. So the question of democratization may not be the question at hand. South Korea allocated billions of won to developing AI. around the same time. It will likely engage in more democratization than China.
It is interesting to see how justice issues, like how facial recognition fails to recognize people that aren’t white men. When you’re training this tech in Chinese data sets, you have a much larger data set – one billion and a half people rather than 350 million (in the US), which allows the possibility to get rid of these biases which offers great potential for global AI for good.
There is also the problem of natural language processing. GPT-3 recently came out, and just like GTP-2, is based on English web pages. This means there is bias from the English-speaking world that is coded in those AI systems. But if you start training those same systems on Chinese, Korean, Japanese, Hindi language pages, you are going to end up with different frameworks. The bigger question will be, is there a way to put these in dialogue? I think this is a much more complicated question. Because there is so much development going around in this part of the world, it opens up the recognition that many of the biases encoded in western development of AI will not be the same as the rest of the world.
Conclusion
In this first part, we introduced the discussion on a global view of AI for good. It includes three main categories: democratizing AI skills, sustainable AI and AI justice. We then framed it within a mustard seed technology perspective. That is, we focus on the margins as opposed to the geo-centers of technological power. We are less interested in Silicon Valley and more on what is happening in the street corners of global cities.
