Mercy Njeri is a health systems specialist. As a consultant in the field, she spends most of her time online researching on health.
Having done that for the past five years, most search engines and social media platforms have studied her Internet behaviour and constantly push health content her way as opposed to other subjects.
“It’s like we’ve become friends. They know what I like and will thus go ahead and prioritise that instead of giving me other things that I am not interested in.”
Ms Njeri’s enhanced user experience is among the many benefits of Artificial Intelligence (AI).
This technology makes it possible for machines or computer systems to learn people’s interests or personalities based on their online activities such as what they search, like, share or comment on while using social media platforms like YouTube, Facebook or Twitter.
The learning process mimics the human brain which becomes intelligent as it is exposed to knowledge or information about a particular person or topic of interest.
AI is already making it possible for machines to learn tasks s that they outperform humans, especially in data analysis.
Companies are using it to understand their clients (target population) so they can offer them services or products that suit their needs or match their interests.
“We are just scratching the surface of AI. But it is clear that in the coming years this technology will control our lives in a big way. So we need to begin having a conversation about its potential benefits and downsides so we can be prepared for it as a country,” said Kathleen Siminyu, data scientist and AI expert at Africa’s Talking Limited.
As the key drivers of AI, major tech giants such as Google, Amazon, Facebook, IBM and Microsoft have begun assessing the complexities of AI early enough as they are aware that its use will largely depend on the buy-in it gets from people and inadvertently governments.
Microsoft recently launched a new book titled The Future Computed: Artificial Intelligence and its role in Society which addresses accuracy, safety and security when using AI.
Somet Kipchilat, technical evangelist for Africa Initiatives at Microsoft notes that because AI is driven by data and companies using AI need to invest in effective control measures to ensure people’s personal information is kept private and protected from theft or misuse.
At a national level, tech specialists are calling on the government to pass the data protection law to guide use of data collected from citizens by companies and individuals.
Mr Kipchilat says since machines develop artificial intelligence by learning from data made available to them, that information needs to be accurate to yield useful decisions.
For instance, a company keen on understanding Kenya’s career patterns can develop a system or app which will use existing data to learn about the sector.
As available data shows, that there are more men in leadership positions compared to women, the system may wrongly conclude that the male gender has superior leadership and managerial qualities.
Using this ‘false’ intelligence, the system may make a decision to push job advertisements for senior management roles to sites with men compared to women.
This will continue to widen the gender gap in employment. Yet, women are just as qualified as men.
There are also concerns that AI is limiting the information that people interact with by giving them more of what they search for online.
Therefore, people supporting a particular political ideology will be swamped with information about it at the expense of other diverse opinions.
This can fuel propaganda, extremism and intolerance to other people’s beliefs.
“Just because I read and liked a couple of online stories today about a particular politician doesn’t necessarily mean that I support him. So I don’t want Google, Facebook or Twitter pushing similar content to me when I go online the next day as is currently happening. That will kill diversity,” said 21-year-old Maureen Anyango, who says she’s tired of machines making decisions for her.
Ms Siminyu states that AI systems arrive at decisions based on data they are presented with.
But when dealing with people, she states that those decisions should be taken with “a pinch of salt” as human beings are complex. How people behave online may be different from who they really are offline.
Rather than relying on AI alone, she recommends that techies need to work with experts in human behaviour (such as those in the field of human rights, ethics or sociology) to enable them make more sense out of the data collected.
Steve Kenei, a data scientist at Development Initiatives says countries like Kenya should prepare well so as to lessen the impact that AI will have on livelihoods as machines take over roles performed by humans leading to massive job losses.
“We can’t fight technology. But we can equip people with the right skills and knowledge that will enable them to transition to other new opportunities that AI will create.”
Nicholas Loki, a Nairobi based AI specialist notes that getting skills in areas such as data science and machine learning —key components of AI — is a challenge in Kenya.
“Most of us have been forced to learn and sharpen our skills online but what we get is not comprehensive. We need our universities and other institutions of higher learning to begin offering these courses.”
Through initiatives such as the Nairobi Women in Machine Learning and Data Science, tech experts are training women on AI to ensure that they will also be part of the AI revolution.
“When Kenyans – both men and women - have the skills, then they can begin investing in AI innovations that address local pressing problems such as hunger, low farm yields, infectious diseases or water and sanitation challenges affecting us,” says Ms Siminyu.
Experts warn that as AI reduces the human workforce, revenues will go to fewer people.
This means that individuals who own AI-driven companies will make all the money in the world.
Countries, therefore, need to think of how such immense financial resources can be amicably distributed to benefit all citizens.