Artificial Intelligence: What Does This Mean For Nigeria And Africa?

Introduction

The term Artificial Intelligence (AI) refers to a simulation of human intelligence that can rationalize and take actions that have the best chance of achieving a specific goal.¹ It can further be defined as an existing network of computer components and connections that mimic the human ability to receive, process, reason and communicate information through various media. It can search through data to combine information for the production of a specific result through multiple media based on an initial request from a human operator. Artificial Intelligence consists of both hardware and software infrastructure that exist in various combinations of each other, depending on the intended use and scale of the AI application. This article will briefly highlight the fundamentals and brief history of AI and assess how best the technology can be exploited for the benefit of Nigeria and Africa.

Brief History of Artificial Intelligence

The early days of AI can be traced to 1950 when Alan Turing first proposed the idea that machines could think in his publication Computing Machinery and Intelligence.² In 1956 the term Artificial Intelligence was first coined by John McCarthy. In 1967 the first computer-based neural network called the Mark 1 Perceptron and was built by Frank Rosenblatt. The 1980's saw the widespread adoption of a backpropagation algorithm which AI used to train itself across different applications. The first AI computer to defeat the world chess champion occurred in 1997. The AI supercomputer called Watson actively participated in the television gameshow 'Jeopardy' and defeated its game champions in 2011. In 2015 a supercomputer with the ability to identify and categorize images with a higher rate of accuracy than the average human was created and used by Baidu in China. The year 2016 saw the advancement of deep neural networks as they powered the AlphaGo program to defeat the world champion of the game 'Go' in a five-game match. What makes this win remarkable is that the world champion was defeated after just 4 moves out of a total number of 14.5 trillion possible moves. In November 2022, Chat GPT was released to the public, and it set the world record of being the first application to sign up one million subscribers within the shortest period ever at 5 days.³

Interestingly, in November 2022, the response of the public to the unveiling of Chat GPT was a variation of excitement, hysteria, and denial. Notwithstanding, the younger generation, specifically the Gen Zs were quick to adopt the technology, especially in view of how instrumental the technology is, as a research tool.

How Does AI Work

Depending on the use case, artificial intelligence requires a large set of data to sift through, learn and train itself to perform the required function. The way artificial intelligence works and operates is dependent on the purpose for which it was created. Notwithstanding its functionality, a common characteristic that exists among all forms of AI is collection of data from which the AI learns to perform the task or tasks for which it was created. In further support of this, the popular chatbot, known as Chat GPT⁴ responded to the question 'What is artificial intelligence?' thus:

AI has a wide range of applications in fields such as healthcare, finance, manufacturing, transportation, and entertainment. It is rapidly evolving and is expected to have a significant impact on society in the coming years."

Types of Artificial Intelligence

AI is typically divided into four categories, namely:⁵

1. Reactive machines: They are designed to respond to specific prompts without the capability of learning anything beyond what it has been taught. It cannot solve a problem it was not trained to solve, and once solved (if at all), it does not store the learning experience in its memory. For instance, IBM's Deep Blue was designed to play the game of chess.
2. Limited memory: this is more complex than the reactive machine. In addition to its training data, this type of AI possesses a memory where it stores records of previous encounters from performing a particular task and compares the records to the current challenge in real time for a possible solution. An example of this is found in how self-driving cars observe other cars on the road for their speed, direction and proximity.⁶
3. Theory of mind: it refers to AI with ability to determine how external factors make it feel and make decisions through a process of self-reflection and determination. This type of AI attempts to simulate the cognitive ability of a human. Whilst some AI models are trained to identify emotional responses in humans, this might sound as though the technology can develop emotional responses, in actual sense, it still remains limited to a mimicked cognitive ability designed to identify a range of physical characteristics and changes to the finest details of triggered pre-set responses and thus perpetuating the perceived capability of emotional intelligence. For now, this type of AI remains at the theoretical stages of development, hence a proven example remains non-existent at this moment.
4. Self-awareness: It is regarded as the AI of the future, where the AI fully depicts the sentient mind of a human being. This AI is said to be conscious of the world around it and it is expected to be capable of understanding what external elements might require of it; not just from what is communicated, but from how it is communicated. Much like the theory of the mind, this type of AI also exists only as a theory and not an existing and publicly proven model.

Artificial Intelligence and the Law

There is a race for supremacy and dominance in the innovation, regulation and adoption of AI among the largest tech companies and the governments of the developed world. In the area of regulation, the European Union (EU) is set to take the global lead in this regard⁷. The EU is at the forefront of enacting the first legislative framework in the world to regulate the use and implementation of AI. The draft negotiating mandate is still advancing through the legislative process at the EU Parliament at the time this article was published, however, the draft is expected to adopt a risk-based approach. The premise upon which the rules were prepared would be by the imposition of compliance obligations on the providers and users of the technology contingent on the degree of risk involved, such as the potential misuse of facial recognition software, deep fakes⁸ and predictive policing.⁹ The implications of the risk posed to people's safety from the technology's application are the primary determinants of how fast or slow AI would be adopted within the EU. This approach is favoured because it is the most accommodating for innovation and subsequent development. Conversely, a major disadvantage would be an underestimation of the identified risk.¹⁰

The United States of America developed a National AI strategy through legislation and Executive orders such as The National Artificial Intelligence Initiative Act 2020 and the National Artificial Intelligence Initiative (NAII) respectively.¹¹ The motivation behind the US government was to ensure its leadership in AI research and development (R&D).The private sector has seen 'big tech' companies such as Microsoft and Google compete in recent months to gain greater market supremacy over the other. In the process, the AI ethics teams have experienced increased degrees of marginalization, to such an extent that Microsoft, Twitch, Twitter among others dissolved their AI ethics team partially or entirely to expedite the deployment process of AI consumer products to the market. Such actions have been criticized by industry stakeholders as short-sighted because the trade-off could be AI consumer products that are unproven and subject to unmitigated consequences far beyond the intention behind their creation.¹²

In October 2022, the US government published a Blueprint for an AI Bill of Rights.¹³ Making automated systems work for the American People. The vision of the government is to create a body of guiding principles through which AI deployment, application and subsequent development are managed in the best way for the American public. The goal of the government is to establish automated systems to power the public sector and all or most government functions, in the administration of public services. The five principles outlined include:

1. Safe and Effective Systems¹⁴
2. Algorithmic Discrimination Protections¹⁵
3. Data Privacy¹⁶
4. Notice and Explanation¹⁷
5. Human Alternatives, Consideration, and Fallback¹⁸

Important to note that the United Kingdom Minister for science and technology addressed the House of Parliament on 29 March 2023 highlighting the goal to becoming a science and technology powerhouse driven by AI by year 2030 through five guiding principles.¹⁹

To read this article in full, please click here.

Footnotes

1. Artificial Intelligence: What It Is and How It is used https://www.investopedia.com/terms/artificial-intelligence-ai.asp accessed 12:00pm, 27 April 2023

2. The History of Artificial Intelligence by Rockwell Anyoha https://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/ accessed April 2023

3. What is Artificial Intelligence? https://www.ibm.com/topics/artificial-intelligence accessed 12:00pm, 27 April 2023

4. Chat Generator Pre-trained Transformer

5. Artificial Intelligence. What is Artificial Intelligence (AI) How Does AI Work? By Alyssa Schroer. 3 March 2023. https://builtin.com/artificial-intelligence accessed 17 May 2023.

6. 4 Types of AI: Getting to know Artificial Intelligence https://www.coursera.org/articles/types-of-ai?isNewUser=true accessed 2:58pm 15 June 2023

7. Europeans Take a Major Step Toward Regulating A.I by Adam Satarino June 14 2023 https://www.nytimes.com/2023/06/14/technology/europe-ai-regulation.html#:~:text=rapidly%20developing%20technology.-,The%20European%20Parliament%2C%20a%20main%20legislative%20branch%20of%20the%20European,
while%20requiring%20makers%20of%20A.I. accessed 4:20pm, 26 June 2024

8. a video of a person in which their face or body has been digitally altered so that they appear to be someone else, typically used maliciously or to spread false information.

9. Prohibit predictive policing and profiling AI systems in law enforcement and criminal justice by European Digital Rights (EDRi) https://edri.org/wp-content/uploads/2022/05/
Prohibit-predictive-and-profiling-AI-systems-in-law-enforcement-and-criminal-justice.pdf accessed

10. Risk Based Approach: Is it Really the Best Decision-Making Approach? https://www.qualityze.com/risk-based-approach-is-it-really-the-best-decision-making-approach/ accessed 11:00am 19 May 2023

11. National Artificial Intelligence Initiative May 2023 https://www.ai.gov/#:~:text=The%20National%20AI%20Initiative%20Act,economic%20prosperity%20and%20national%20security accessed at 9:50am 27 June 2023

12. As AI booms, tech firms are laying off their ethicists by Gerrit De Vynck and Will Oremus 30 March, 2023 https://www.washingtonpost.com/technology/2023/03/30/tech-companies-cut-ai-ethics/ accessed 2:00 pm 25 May 2023.

13. BLUEPRINT FOR AN AI BILL OF RIGHTS: MAKING AUTOMATED SYSTEMS WORK FOR THE AMERICAN PEOPLE- The White House Published October 2022.

14. Ibid. Automated systems should be developed appropriately with relevant data. The system must be tried, proven and tested to be trusted to produce the desired result upon subsequent use. Users or subscribers must not be subjected to a system that is fundamentally ineffective and irrelevant. AI ethicists are the professionals directly tasked with safeguarding this principle but the private sector are either downsizing or currently dissolving their AI ethics teams to deploy their consumer products to the market, to avoid losing market share.

15. It is established as the protection of users of automated systems from unjustified treatment premised on any classification whatsoever.

16. Data collection forms and subsequent protection from violations of privacy through strict use of data within the specified parameters of use.

17. Every automated system is required to be described in the simplest language for users and other stakeholders to fully comprehend the scope of the automated system.

18. A mechanism through which users of automated systems can seek redress from human interfaces in the event the wrong of inappropriate solution is generated.

19. A pro-innovation approach to AI regulation, by UK Govt last updated 22 June 2023 https://www.gov.uk/government/publications/ai-regulation-a-pro-innovation-approach/white-paper#ministerial-foreword accessed 1:00pm 30 June 2023

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.