Retaining superior intelligence: The role of regulators

By Babatunde Olafisoye

I have often wondered which would be superior in the coming years, Human Intelligence or Artificial Intelligence, and a key contemplation on this subject is which would take control over the other? We have heard experts repeatedly talk about how Artificial Intelligence has significantly been integrated into our daily lives, and any attempt by us to thwart this integration, if at all remotely possible, would be akin to reversion to the dark ages again.

To give us a glimpse of what we could expect, a lot of movies have stretched our imagination on the extent to which this technology is going to affect everything and anything. A recent movie that is quite related to this subject is the Hollywood blockbuster; Godzilla Vs Kong. The movie which was directed by Adam Wingard illustrates a battle for supremacy between two Alpha mythical beings called “Titans”. A group called APEX in a bid to prove that humans must remain superior created a giant robot (Mechagodzilla) to battle these titans and we know how the robot fared in the end.

The boggling questions then are; are we superior? If yes, can we stay superior? If no, how can we come back on top? In all these, a critical role amongst key stakeholders that must not be overlooked and cannot be overstressed is the role of Regulators.

This piece critically discusses the concepts of intelligence vis-a-vis the meanings of both artificial and human intelligence, the interconnection that underlies their functionality in the general scale of our existence, the grapple for supremacy and what it portends for the future, as well as the indispensable role of regulators in finding a workable balance between them.

What is Intelligence?

Biologically speaking, a narrow and simplistic way to define Intelligence is that it is the ability to think, to learn from experience, to solve problems, and adapt to new situations. Think about our world and the several unique creatures in it. We can agree, unbiased, that humans are the most complexly intelligent creatures currently roaming this planet. We have even figured out a way to measure our intelligence, including that of other animals just to prove that we are more intelligent. There are significant gaps, in terms of intelligence and complexity between the brains of a worm, a chicken, an ape, and an adult human. However, in recent times we see a potential contender — Artificial Intelligence (AI).

What is Artificial Intelligence (AI)?

AI just like every concept is devoid of a uniform definition. However, in simple terms, Investopedia concisely expressed it to mean “the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions.” Let’s agree for our purpose that “Intelligence” is purely about information processing and the faster and more efficiently an entity can process information to give a precise result, the more intelligent that entity is.

An undeniable attribute of this human creation called Artificial Intelligence is that it is already performing more highly efficient and accurate specific operations than humans can perform, and we are certain that these specific creations, pieced together, will very soon surpass human intelligence. We simply are working on the tools that will create a foundation for the (true, “Super Intelligent”) AI.

For a comprehensive understanding, in terms of complexity, there are different forms of AI.

These different forms are either Strong AI or Weak AI. The results of creating software and algorithms under these categories have brought about categorising AI as Reactive Machines, Limited Memory, Theory of Mind, Self-aware, Artificial Narrow Intelligence (where we currently are), Artificial General Intelligence, and Artificial Superintelligence.

I will not argue whether or not we are at the brink of an “Age of Ultron” scenario, I would however agree that we are slowly reaching the level of creating such an intricately mesmerising entity. With Machine Learning at its full potential, we will witness AI on a seamless autopilot. But before we get there, our imaginations can only run wild on the extra opium fed to it by Hollywood Thriller movies.

Limiting our experience to what we have today, mostly Artificial Narrow Intelligence, we see our computers increasingly perform simple tasks better than we normally would. Take for instance your Google Maps, it practically knows every street name, how to get there and approximately how long it would take to get there. Relying on a similar information database are self-driving cars and semi-autonomous drones, these technologies have in some cases proven to make better judgements than their human counterparts.

We must not fail to mention AlphaGo — the highlight of AI development history. AlphaGo, a Google DeepMind project is an AI agent that is specialized to play Go (a Chinese strategy board game) against human competitors. In a test, after 72 hours of supervised learning, the AI beat the human world champion, Kie Jie. The next even more significant move was the creation of AlphaGo Zero. This AI was capable of teaching itself to play the game and after much practise against itself, beat its predecessor.

Another significant example of this type of AI is Sophia. Sophia is a realistic humanoid robot capable of displaying human-like expressions and interacting with people. While a few writers have oversimplified its existence to be like a chatbot with a face, it truly is a wonder to observe. It is even a citizen of Saudi Arabia. I like to call it the marriage between AI and Robotics [maybe a pre-evolved Ultron].

Perhaps the most significant development in this AI form is OpenAI’s GPT-3. GPT-3 is a creation of OpenAI, a research business co-founded by Elon Musk and has been described as the most important and useful advance in AI for years. This AI is capable of generating an essay in a few seconds, answering any question, generating music, creating designs, and building software components, translating to and from a variety of languages, knows billions of words, and is even capable of coding! Because of all the data GPT-3 has at hand, it requires no further training to fulfil language tasks. Put simply; it’s an AI that is better at creating content that has a language structure – human or machine language – than anything that has come before it.

The Risks and Control Mechanisms

Stephen Hawking and Elon Musk share similar thoughts on Superintelligent AI being the greatest feat man could ever attain, however, it could be the ultimate cause of our extinction if not properly checked and managed.

As Elon Musk puts it, “AI will be the best or worst thing ever for humanity.” The reassuring thing about the level of AI development we have today is that it is currently not capable of engineering itself and it is nowhere near human intelligence. And while we are still far away from what would potentially evolve into Artificial Super Intelligence, what we have today is only narrow, can only process huge amounts of data and provide results on specific tasks with little or no human intervention.

Consequently, I believe the technology is at its incubation. It is then imperative that this technology along with its infinite possibilities be properly understood and in the best way, regulated so as to prevent the risks of a rogue or runaway Super Intelligence. In other words, control it from its infancy.

On preventing the technology from eventually causing our extinction, several solutions have been theorised. One of such solutions is to isolate the technology and keep it confined in a “box”. This could mean isolating it from accessing the internet, limiting its contact with the outside world. The problem this would inevitably birth is a great reduction of its ability to perform the functions for which it was created.

Another solution is to design a “theoretical containment algorithm” to ensure that an artificial intelligence “cannot harm people under any circumstances.” However, an analysis of the current computing paradigm showed that no such singular algorithm can be created. Another one that is radical but seems to be working is implanting this technology in human brains. This will facilitate an efficient control of the technology.

Regulations and the Roles of Regulators

Whichever way we decide to solve this problem, one key aspect must not be overlooked — the role of Regulators. We have seen good attempts by various jurisdictions to keep up with regulating the emerging trends in technology and more particularly, Artificial Intelligence. Some countries have managed to develop regulations and guidelines targeted towards ensuring that the wide haze of AI is kept under close watch while still adapting it to meet the crucial and ever-expanding human needs as the world dives into the future.

In June 2018 the European Commission set up the independent High-Level Expert Group on AI to provide guidelines on how AI can achieve a unique degree of reliability.

In light of the guidelines, the Commission has since published its own ‘White Paper on Artificial Intelligence – A European approach to excellence and trust – on 19 February 2020. The Commission’s White Paper outlined that the most alarming risks to be addressed concerning AI are those it poses to the landscape of fundamental rights, the privacy of data, safety and effective performance, and liability identification. The Commission insists that the optimal method of regulation should be risk-based to ensure that reactions to AI development are commensurate with and do not stifle innovation.

In place of proposing regulations at this level, the Commission has laid out some legal requirements which must be captured by any regulatory framework to ensure that AI continues to be dependable, and subjective to the values and principles of the European Union.

Following this White Paper, the EU recently projected that its draft regulation will be released on 21 April 2021.

On 21 April 2021, the European Commission adopted a proposal for a regulation of AI Systems, one which it describes as “the first-ever legal framework on AI.” The AI Regulation will impose significant obligations impacting businesses across many, if not all, sectors of the economy. The AI Regulation will prove controversial, touching off a legislative battle lasting at least until 2022.