Artificial Intelligence (AI), a popular topic of discussion at dinner parties nowadays. There is much speculation in the media, often negative, about what it might mean for our future. This is unusual, given how widespread adoption of new technologies is in our everyday lives. What is the reality of AI today and what are the myths, if any? In this article, I am going to ask ourselves to use our human intelligence to think about AI, what it really is, what are the myths surrounding it, whether it poses certain dangers and risks to humans, and (if there are dangers) how we might be able to protect ourselves.
I am not an AI expert but the growth of AI parallels my working life as a scientist and technologist. Over the last 45 years, I have participated in the fast-moving journey of the development of computing technology and its ubiquitous deployment across the world. But, just like the negative impact industrialization has had on the world's climate, many, including myself, are beginning to realize how blindly we have been allowing ourselves to be dependent on technology. But don't worry, this isn't going to be very technical as I will raise several questions on moral philosophy and ethics -- areas that are beyond the scope of science and technology.
The background
Jonathan Sacks, a well-known theologian in the UK, once said, “Science takes things apart to see how they work. Religion puts things together to see what they mean.” Of course, he meant that it is us humans who have been using science and religion in that manner.
So, how have been using science and technology prior to the arrival of AI? From cars, airplanes, TVs, computers, and smartphones, science has spawned a vast array of technologies, and we humans have become addicted to them. So much so that our economies will fail to survive without the development of new technologies. Our daily lives are, therefore, strongly intertwined with and entrapped by technology. Meanwhile, religion, which for centuries had provided a strong moral and ethical counterbalance in our lives, is losing its hold. This is particularly so, in the technologically developed world. If we then allow AI to fully take over our lives, does the blame not lie with us for allowing humanity to lose the sense of meaning and purpose of life?
What is AI?
Seventy years after Alan Turing wrote his paper entitled Computing Machinery and Intelligence and 50 years after Marvin Minsky invented artificial neural networks, today's AI has been termed by some as the most profound change in human development. So, what is this AI? AI is a computing machine with artificial neural networks, much like those in our brains, which allow the machine to learn, when we feed it with the outputs of intelligent humans. Such a machine is, therefore, both artificial and intelligent. When attached to large datasets, the memory of these machines can be bigger than that of individual brains, and their outputs more than that of individual humans.
Today's AI
AI today is not only able to deal with single human objectives, such as driving a car, translating languages or human speech, it can be better than single human experts in, for example, identifying complex new diseases, drug development for pandemics, and winning chess games. Today's AI is also used in internet search engines and various smartphone apps. They can learn our individual preferences, biases, and decision-making and can predict individual human behaviour. There is now increasing evidence that some governments and corporations have been using AI to subtly manipulate our behaviour for their private gain.
What is ChatGPT?
ChatGPT is an AI chatbot developed by OpenAI. This is a company in which Elon Musk is involved and in which Microsoft has invested. Built on natural language models, ChatGPT can produce impeccable human writing. Hence, there is a danger that these chatbots may flood our information channels with propaganda and untruth. The concern is that humans might not be able to detect and avoid the possible faking, impersonating, and lying using AI.
The threats of AI
An area of strong ethical concern is the use of autonomous and self-guiding weapons. When a human kills and/or watches another human die, it leaves a heavy psychological impact that has a deterrent effect. Remote killing via machines, on the other hand, has little or no deterrence effect. There is a danger that AI might be used to produce large numbers of cheap weapons of mass destruction that could lead to genocides. The UN, EU, China, and 30 other states representing the majority of the world's people, are for a ban on self-guided weapons. The UK, US, Australia, Israel, and Russia are not in favour. A ban on eliminating swarms of small anti-personnel weapons is supported by the Red Cross. An area of strong social concern is the use of AI in economics. On the positive side, AI can be used to augment current jobs by meeting unmet needs which are too difficult for humans. But there is a danger that AI may end all repetitive jobs, thereby most full-time jobs. Aristotle is known to have said that full mechanization will be the end of work. Therefore, there is a danger that this may lead to the loss of most of our jobs, including the fulfilling ones.
The rollout of AI is starting to make us think very carefully about human objectives and the value of human lives. Younger scientists like Will Macaskill of Oxford University are introspecting what makes a good future, or “longtermism.” This is positive, given that, in the past, we humans have been minimally concerned about the impact of human actions and of the future of human objectives. Increasingly, scientists are pondering about what the use of AI in economics means and whether future generations will lose the incentive to learn or do anything, not have the basic skills to live life to the full, not need to strive or to contribute to society, lose their autonomy to machines, and effectively live as useless zombies.
Artificial general intelligence
Thus far, I have been talking about special purpose AI, which can achieve specific objectives set for them. What about general-purpose AI, or Artificial General Intelligence (AGI)? An AGI is a highly intelligent computer that learns and develops autonomously and understands its environment without the need for supervision. AGIs do not yet exist. Although there is exponential growth in AI capabilities, this is happening within an extremely narrow domain. Such growth cannot be extrapolated to mimic or replicate the cognitive capabilities of human intelligence. In fact, there is evidence that the current AI paradigm is rapidly approaching its own internal limits.
That -- while ground-breaking advances will continue happening -- these will be capped by the fundamental constraints of this paradigm and AGI may never exist. Take for example, ChatGPT. It will only improve incrementally as a natural language processor, rather than produce a fundamental breakthrough into a higher-order intelligence. That's why Open AI is not working on a fifth generation of ChatGPT.
The difference between human and machine intelligence
Let us briefly consider the two words, “artificial” and “intelligence.” Artificial means not arising from natural causes, or not a living thing. We all know that science has not found a way to create life from basic chemicals and minerals. Not the cells or genes, nor their reproduction, evolution, or management thereof. An AI machine is no comparison to a living human, from a biological, chemical, or physical perspective, although it is possible to make an AI robot look like a human from a distance. Humans are far superior as a complex living system. In terms of intelligence, the AI machine learning models are built on the principle of induction. This is the inferring of patterns from specific observations or, more generally, acquiring knowledge from experience. Induction uses big-data processing -- the more the observations, the better the model. We feed an algorithm with thousands of labelled pictures of cats, or have it play millions of games of chess, and it correlates which relationships among the input result in the best prediction accuracy.
Some models are faster than others, or more sophisticated in their pattern recognition, but at the end they are all doing the same thing: Statistical generalization from observations. But human intelligence is based on an additional and far more creative approach to generating knowledge, called abduction. Abductive inference allows us to creatively select and test hypotheses, quickly eliminate the ones which are proven wrong, and create new ones as we go along, before reaching a reliable conclusion. This allows us to generalize about the world, even about things where we have no direct experience. This is where some of our best ideas come from.
The myths of AI
In his book entitled The Myth of Artificial Intelligence, Eric J Larson warns that the claims of researchers being able to produce an AGI in the next two decades is based on a narrow and reductionist model of intelligence and consciousness, and are therefore flawed. He believes there is no good theory of how abductive inference works in the human mind, and no one has any idea how to recreate abductive inference for AI. So, how could AI lead to higher than human intelligence?
What about human emotions and consciousness?
So far, I have been talking about intelligence, inferring through deduction, induction, or abduction. But humans also have feelings, moods, and emotions. How can we build machines that will evolve like humans did, with emotions, if we don't understand how evolution works in any detail? Neither do we have any definitive way to create or detect consciousness, even in human beings. So, how likely is it that AI will develop consciousness?
Is AI an existential threat to us?
To conclude about the dangers of AGI, therefore, the idea that AGI poses an existential risk to humanity is not one that is grounded in AI innovations to date. The real risk from AI is not from superintelligence, it is from the highly effective, highly scalable manipulation of human behaviour that AI enables, and its malicious use by corporations and governments.
Societal choices
How AI affects industries and sectors will ultimately depend on societal choices. If we choose to roll out and apply AI within the current hierarchical, social, and economic order, we could see it amplifying and reinforcing prevailing inequalities, injustices, and social and geo-political divisions. The question is whether AI could lead to a breakdown of society, a new dark age, or even the collapse of human civilization.
Positive alternatives
We could mobilize AI to unleash unprecedented productivity and creativity. We could distribute access to information in new and powerful ways that create new benefits for people at all levels. AI could be used to augment our collective intelligence as a species, and to help us renew and revitalize human civilization.
Controlling AI
Regulating AI is going to be fraught with difficulty as it is not at all easy to predict what is good for our future. The speed of AI development is far outpacing our ability to implement its ethical underpinnings. Nonetheless, regulation based on products could be used to stop the use of AI to impersonate humans and deepfakes. Regulations based on human rights could be used to preserve the right of human mental integrity. It is a good thing that people are more sceptical of technology nowadays, eg of algorithms refusing job applications.
Regulation by itself cannot control things. We will need widespread training on ethics. Children will need to be taught how AI should or shouldn't be used to benefit societies. Teaching adults will, however, be more difficult.
Summary
If we are to obtain the superb benefits of AI without losing it all, we will need to do some very hard thinking about our long-term objectives, including the future we wish for our children and our children's children. The biggest challenge will be that of time. The evolution of AI is far exceeding our ability to control it using usual means. We could find ourselves unable to control it if we don't act concertedly and in a united manner.
I will end with a quote from TS Eliot: “Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information?”
Suheil Shahryar is a retired IT professional based in the UK.
