Encoding values and culture in AI
How can AI reinforce African values and culture?
If someone snarled an insult at you in English, you would continue with your business. Until the same insult is repeated in your vernacular. For luos, there’s a difference between ‘skinny’ and randere. Think of other translations in your languages, and you’ll discover that we lose meaning every time a word is translate into a language that we were not raised in. Well, the same way we lose our society values when we miss the opportunities to take part in the global information society.
The opposite development trajectories between Africa and Asia are so remarkable. In the 1950s, an African used to earn twice as much as an Asian: now it is the other way round. A friend of mine rightly argues that it is because of the different styles of colonization that took place between these two regions. Africa was washed of every culture and values it held, while the Asians firmly held to theirs to date. Majority of African communities abandoned their religious beliefs for Christianity, which was brought by the colonial masters. What became of the mugumo tree? Lwanda magere? And the other supernaturals we believed in? Our colonial masters told us that to pray facing a tree was backward thinking, and we dropped it like it was hot. To date Asians still have Budhism as its dominant religion.
The little that has remained
Majority of the unconnected in Africa are the older folks. The Kenyan population that is connected to the Internet and use other digital services range from ages 15–50 years. A paltry fraction of those above 50years use digital services, except for the solar and mobile money. Note that this is the population that has lived during the colonial and post colonial era. It is important for us to make efforts to include them in the digital spaces. If we are not careful, this population is going to fade away with so much content.
A bit about Artificial Intelligence
Artificial Intelligence (AI) is a software that autonomously develops its own way of thinking and updates itself, in super-normal speeds. What powers and directs its thinking is the data that is fed into it. For example, a computer may be fed with data of events and multiple variables. Let’s say Apondo’s farm is 3 hectares, soil property 4% nitrogen, Climate conditions-annual rainfall of 123mm/year and it produced a yield of 100 bags of grain. 1000 of such data will be fed into a computer such that it will be able to carry out regression analysis, and come up with an algorithm based (based on these variables) to predict farm yields with using different input and environmental conditions.
Where y=farm output x1=size of the farm x2=soil property and so on.
AI is designed in such a way that we don’t realize that the computer is deciding for us. That’s why many of us talk of it as technology coming soon, yet we interact with it daily. The one tap response we use for our google mail is AI, what we see on our Facebook TL is determined by AI and, if you haven’t noticed yet, sometimes you type a word and your Samsung keypad continues suggesting the subsequent words based on your typing patterns –that too is AI. Insurance companies will soon start charging your premiums based on variables such as what you bought using your mobile money, weight, age, etc. That is AI. AI has permeated into every sector of our economy: from healthcare, our digital social interactions, politics , entertainment, mobile money and, to banking.
Africans, just like the Asians are of high context culture. Words bears so much symbolic meanings. The cohesion in the communities determines peoples, lifestyles, from political, social and economic fronts. From these social structures, You will find out the meaning of food in ceremonies and in ordinary days. These values determine our perception of life, our interactions and how we put food on our table. Now imagine our levels of productivity when the AI we use is fed on such data.
Just like public policies, tech innovation thrives when it aligns to cultural contexts and good strategies. Tech policies are encoded in their design, just when the tools are made. Technology is therefore not neutral as we have previously assumed. Computer languages and codes are designed by humans in their different contexts, and computers eventually function within the limits of the commands they have been fed --by humans.
So far, the AI we use is largely fed foreign data that don’t relate to us. You can imagine the loses that will be made when farmers use such AI products, or how much we will lose as a community when we are not keen in developing policies that ensure our values are reinforced as well.
The policies we need
1. We need to develop enabling policies for local based AI initiatives. Our data protection policies should not be too restrictive such that they discourage The SMEs and startups from using local data to build AI-based solutions. TAX incentives should be given to such entities.
2. African governments should be proactive in publishing open data. The first step would for Kenya be to revive the open data initiative that was launched with so much hype.
3. Review local content policies so that our older population is able to use technology, or at least generate content that may be used to make decisions that are specific to our problems, and are also based to how we are organized as communities.
4. The blockchain and AI task-force should come up with our specific standards of accountability and reliability.