Is the future of AI inclusive?
“The idea is that we are building our own MIT here,” says Roya Mahboob, founder of Digital Citizen fund, and one of the first Afghan women to become CEO of a tech company in the country.
The project she is referring to is something she has envisioned since she began her work. Ada-AI, a newly launched nonprofit working to ensure that artificial intelligence (AI) and other emerging technologies are developed and adopted in inclusive and equitable ways, is working with Mahboob to make that vision a reality. The organization recently launched a fundraising campaign to open a STEM school in Herat, Afghanistan.
Organizations like Ada-AI are key to ensure a more inclusive future in this white male-dominated industry.
An industry dominated by young white males
Doctor Virginia Dignum from Delft University, who is working on creating models that ensure that AI systems behave according to a set of ethical parameters, is keen to highlight that the way AI systems are being developed needs to have a strong “human” perspective, rather than solely an economic one. The way algorithms are fed, she says, should be adaptable to other cultures.
Although the bias is not intended, say Doctor Dignum, it is a fact that most AI work is done by white young males and only a small portion of the population is involved in producing AI algorithms and responsible for organizing and using the data that is available.
Her work focuses on making sure that in the future there will be more representation of cultures, genders, disabilities, and races in AI growth.
A recent report by Oxfam shows the current state of inequality globally, where last year, 1 per cent of the global population earned 82 per cent of wealth generated. With an already existing digital divide, the risks are that AI could make this worse.
In terms of AI, Porter says you could almost draw a parallel with a child in terms of the cognitive stage, and, like a child, AI can pick up existing biases from the people who manoeuvre them.
Examples of this are numerous. A report by ProPublica, for example, showed this when a software used to predict future criminals predicted that 61% of these criminals would be black. When a white man and a black man were incarcerated, a program predicted that the black prisoner was more likely to commit a future crime, despite the criminal history of the white man being more serious. The machine was proved wrong later on.
Existing inequalities in access to healthcare, as well as education, could also be exacerbated. The same goes for existing inequalities between wealthy and poorer countries.
Porter says that if AI learns in a non-inclusive way, the equal treatment of people will be more in jeopardy than it is now. A non-inclusive growth of AI technology could destabilize society through rapid automation and have ripple effect consequences that could exacerbate conflicts and inequality.
Tech in developing countries
Andrew Schroeder, Founder of WeRobotics, an organization focusing on creating sustainable tech projects in developing countries, is optimistic, saying that we can “build new conversations off of (this) history” and learn from it.
During his experience working in the development sector, and in tech, Schroeder notes that many tech projects that were done in developing countries had very little continuity after the project was finished.
A common thought would be that the capacity to develop the technology is limited in developing countries, but Schroeder’s work at WeRobotics proves that there is a capable and experienced force that can participate in AI development and implementation in developing countries.
For example, one of WeRobotics ‘Flying Labs’, local innovations centres that aim at accelerating and scaling the impact of development projects locally, is being led by a local force that is currently in the process of making the centre a Tanzanian entity recognized by the government.
This proves that the potential is there, and that “it could be just that the business model is not developed yet,” says Schroeder.
Afghanistan as a leading example
Mahboob supports this vision and wants to bring her country to the forefront of the global discourse on AI. Just a year ago, Mahboob, who sponsored an Afghan girls’ Robotic team to compete in the US, found herself in a difficult position when the girls were not allowed a visa to US under the Trump travel ban.
The girls were eventually given a visa after the fact caused outrage in the media, but it did more than that. As the girls gained their “celebrity status,”as Porter says, experts in tech and AI business saw this as a blatant example of non-inclusion in AI for the future.
“It’s just about giving [them] the right tools and education”
The girl’s team recently won a competition at the Estonian Robotics Festival, says Mahboob, for building a prototype of an agriculture robot using solar energy to support small-scale farmers.
“They proved that it’s just about giving [them] the right tools and education” she continues, pointing out that after they won the competition in Europe they returned to find the support of the Afghan President and their community.
“(AI) will be the future, and I want Afghanistan to be part of it” adds Mahboob. For Ada-AI, the launch of the school project is the first stepping stone, but they hope to continue their work in creating bodies that can regulate this growth.
“The future of AI will be as inclusive as we make it”
Schroeder agrees, saying it’s really important not to only think about development and implementation of the technology, but to think of all the processes around it, from access to the technology to the setting up of new institutional models able to build tangible projects, to inclusion of communities in decision making.
The challenges are not scarce. The biggest challenge, Porter says “Is in countries where we’re battling with culture change” for example in countries where girls’ access to education is limited, like Afghanistan, “and where families risk being ostracised from the community for allowing their daughters go to school.”
People like Schroeder, Porter, Dignum and Mahboob are starting fundamental conversations which will write our future.
There is a possibility to reach more stakeholders, Porter says that her organization has until now had a lot of support from white males, who are very aware of the risks.
“The future of AI will be as inclusive as we make it” concludes Schroeder.