On one of my social media tours exploring content on Facebook one afternoon, I came across a post by Prof. Fred McBagonluri, President of the Academic City University, confirming the accreditation and roll out of BSc. Artificial Intelligence (AI) at the institution. Seeing that post got me excited and within a couple of minutes, I shared the good news on my LinkedIn and Twitter profiles. I saw positive comments and likes from both the AI fraternity in Ghana and abroad. Indeed, I could sense the excitement from people and how delighted they were to see this progress and step in the right direction. While I kept reading the comments and responding to personal messages from well-wishers, I began asking myself again, “Is my beloved country, Ghana, ready for this? Do we have the necessary labs to facilitate the introduction of a new program aside from our traditional programs?” Hence, I decided to write this article to spark the conversation on AI.

Let me first explain what I think AI is; artificial from the Oxford dictionary means man-made, whereas intelligence is the ability to acquire knowledge and skills. Hence, coining these two terms, AI simply means the ability to convey man-made knowledge or skills to a machine or computer. In essence, computers can solve tasks that are abstract for humans with so much ease. A good case example is the defeat of Garry Kasparov by IBM’s Deep Blue chess-playing system in 1997. Typically, humans acquire much information about the world through a subjective and intuitive manner which is hard to explain more formally. Thus, one of the major difficulties in artificial intelligence is figuring out how to transfer this unstructured knowledge into a machine.

In spite of everything, we have all come into contact with AI deployed solutions in our everyday life at some point in time. These range from using google maps, to navigating our way through unknown destinations to social media sending you recommended items based on previous items you either purchased or viewed. Recently Open-AI, a firm based in San-Francisco launched its new technological language model that is capable of producing human-like text, answering questions, completing codes, and also summarizing long text. Also, Apple Inc. uses AI in nearly all features of the iPhone. Its facial recognition systems enable users to add extra security features and these are all developed using computer vision models (a subdomain of AI). Facebook, Google, and Snapchat also use these computer vision architectures to enable one to identify photos of people and then tag photos along with the name. The COVID-19 pandemic also led to the development of further uses of AI systems in the healthcare space. Systems used in tracking, monitoring, and screening COVID-19 patients were implemented, which greatly helped countries manage the pandemic. Research conducted by two IBM researchers strongly suggests that the use of AI could reduce the number and time spent in conducting clinical trials by more than 70%.

The above examples re-emphasize the role AI plays in our lives and how much more transformation and satisfaction it could bring. Nonetheless, a survey conducted by Smith and Anderson sought to find out views from 1,896 experts on whether AI will displace more jobs than they will have created by 2025. They found out that 48% of these experts believed AI will displace more jobs than it would have created by 2025 whereas the other 52% believed that it will not displace more jobs than it creates. I concur with the latter and believe that AI will create more jobs. Humans need to be future-ready for this possible revolution.

The dawn of the Industrial Revolution has brought forward new jobs through human ingenuity and the revolution of AI in the 4th Industrial Revolution will be no different. Malone et al. recently suggested that AI will enable new industries to emerge thereby creating more jobs. This has already been proven by the demand for AI skills measured by the increasing number of posted vacancies (Alekseeva et al., 2020)

In this regard, it is crucial for educational institutions, government and industry players to effectively collaborate to pursue the AI agenda for the socio-economic development of Ghana. To achieve this, there is the need to ensure that the next generation is well prepared and ready to take over the new jobs that will be crated through AI.

The introduction of BSc AI by Academic City University is absolutely a step in the right direction and I am convinced that many higher educational institutions in the country will take a cue from this novelty. I am a strong advocate of Academic City University’s values and mission. This is not only because I lecture at the university but rather stems from the University’s unique way of training students which allows students not only to become consumers of theoretical knowledge but also practitioners of what they learn. We need more institutions of this nature if we want to reap the benefits of AI.

Subsequently, the recent advances in the field have brought extraordinary abilities of AI being able to replace human decisions at critical points such as loan approval systems, job hiring, bailing, etc. However, AI systems learn patterns from historical data and are then able to make decisions in the future. As such, historical data have a significant role in determining potential hired candidates in the future. But, what happens if the historical data in itself is biased? The AI systems developed will only further amplify the bias in the data. There has been recent news that indicates how AI systems have discriminated against people due to their gender, race, or ethnicity. Researchers have proposed the non-use of sensitive attributes to train AI models. Regardless of this deficiency, this does not completely eradicate the issue of bias as AI systems could still find proxies for sensitive attributes. That is, one’s zip code could be used as a proxy for an individual’s ethnicity or race (Tal Z, 2014). Amazon developed a recruitment tool that turned out to be biased against women and eliminated all resumes belonging to women (Hamilton, 2018). Also, the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) which was also used in the US to help judges make decisions on pre-trial bail turned out to be biased against black people (Larson et al., 2016).

All these issues, therefore, point us to the fact that, as we welcome the potential benefits AI can bring to us as a country, special attention needs to be paid to ensuring that these AI systems developed are fair and unbiased. In doing this, there is a call for action for all parties; AI engineers, academic institutions, governments, etc. We need to first acknowledge and admit that bias in AI is indeed a huge problem that needs to be addressed. Following that, a chunk of the issues in AI bias arises from the training data. AI systems should not be built using any cheaply available data without proper consultants in investigating the source of the data, how the data was collected, etc. Important questions should be raised and a thorough investigation done before any data is used in developing AI systems. Amegadzie et al., 2021 throws more details on how to leverage the use of AI for the betterment of the good citizens of Ghana.

This implies that both government and other non-governmental organizations should invest a chunk of their resources into proper and acceptable data collection procedures. In addition, governments can use different verification and audit tools to assess the risk and impacts of any ready-to-deploy AI tool to ascertain that it complies with the AI ethical guidelines. Universities will also need to introduce AI ethics courses as part of their AI curriculum. This will ensure that students are well-grounded about the ethical implications of any system developed. If all these measures are put in place, we can be confident that the next generation of AI engineers and developers will enjoy all the numerous benefits of this emerging technology and not use it as a mass destruction tool that affects marginalised groups.