AI will never be the same as humans

 

From Mohammad Tariqur Rahman

Only two decades ago, artificial general intelligence (AGI) was no less than science fiction. Large language models (LLM) can now generate “tools” or “intelligent machines” that are more than just driving cars without going for a driving lesson or acting based on prediction or extended imagination.

Not to mention, people enjoy meaningful conversations with intelligent, empathic AI, which can express hopes and dreams and help people work through difficult emotional challenges.

It’s no surprise that people, who do not know how an AI is built or how an AI works, are pounded with questions in their minds. Can AI solve the climate crisis, predict the emergence of another pandemic, diagnose diseases like cancer early, predict famine in Africa, or give a solution to our quest for education for all? Or, will AI determine whether China or the US will dominate the world?

While we are asking those questions, we have already sensitised to the possibility of AI replacing many more human jobs in the near future than we are predicting today. Nevertheless, the biggest question comes to our mind: will AI one day start to control human beings? Or will there be a doomsday to shift global control from human to AI?

To many AI scientists and technologists, the answers to those questions depend on how humans would let AI grow and develop.

To Mustafa Suleyman (a British AI entrepreneur, the CEO of Microsoft AI, and the co-founder and former head of applied AI at DeepMind) AI is more than a tool for “Homo technologicus”. Mustafa used the metaphor “digital species” to define AI, who will be more than a mechanistic assistant but will act as a confidant, friend, or colleague.

Wondering why would we choose an AI as our companion? Mustafa justified that an AI can consume more than eight trillion words in a single training compared to only 8 billion words that we consume if we read 24 hours a day for our entire life. While an average man can hardly remember one-tenth of those words, an AI will remember it all.

All in all, an AI has the enormous potential to have an almost perfect IQ and exceptional EQ.

This is exactly why we would want to have an AI as an intelligent companion who would provide the services of a lawyer, financial adviser, doctor, designer, and planner at the same time.

Then, Mustafa defied his own metaphor and went beyond to define AI as a “digital species” – simply because AI is not just another invention of Homo technologicus but AI itself is an infinite inventor.

To him, every bit of how we have evolved, changed, and created around us in our human world today – AI is not something outside of it. Rather it is the very opposite – it is the whole of everything that we have created. It is us.

Does it mean that AI will be the same as Homo sapiens? In many aspects of human attributes – it would – as Mustafa and many more AI technologists prefer to say so.

AI would not debate on who their creator is. On the contrary, humans build their worldviews on faith as believers or as atheists. They also change their worldview from an atheist to a believer and vice versa. But an AI will not need to debate on what worldview they would belong to. They will know their creator – Homo sapiens – unless they want to deny that.

Perhaps an AI would accurately predict if a Democrat or a Liberal would win a presidential race. But it would not debate on whether to support the Democratic candidate or the Liberal candidate based on a political worldview.

It may not be far from today that AI will generate original creative ideas without being explicitly trained on them. Hence, an AI would not only recreate Mona Lisa or a Weeping Woman, but also create some new original art. However, unlike an AI, Vinci or Picasso did not need any pre-coded programming to translate their skills and talents on the canvas.

Unlike us, an AI would not have to find an anti-ageing solution to enjoy immortality. No doubt materials wear and tear. But recreating the same AI in a nearly immortal material or semi-material body would take a few seconds or even shorter than that.

Again, an AI mother would not have to wait nine months to deliver a weak feeble baby as the next generation. In fact, an AI will be able to recreate itself or perhaps an advanced version of it in nine nanoseconds or less, granted its creators permit self-development, though not through reproduction.

Albeit a next-generation AI would be born with enormous intelligence and information coded into its memory. That would make them super-intelligent but not genius. In contrast, a weak feeble human baby may show talents like writing poems, drawing pictures, memorising thousands of words, or calculating complex math without any pre-coded programme in their brain. That makes a human baby naturally genius without a code beyond human comprehension.

We see an AI providing love and care in old homes, but an “old” AI may not need love and care after thousands of years of service at old AI homes. Humans have given immortality to an AI without knowing fully how the presence or absence of a “soul” can make a human die.

No matter how we want to define AI – as a tool, an intelligent machine, an intelligent companion, digital species, or even as us – an AI will never be the same as human beings.

Should we then treat AI more than a technological advancement? No doubt that most of mankind has no slightest idea of how an AI will work in the near future – but some of us know it well.

Surely, it is not an AI that is forcing us to develop the next generation of AI that could one day take control of human beings. On the other hand, it is some of us who have the talents and genius that can make a self-creating, self-programming AI that will technically be more powerful than humans.

Some of us might think that making powerful AI is the future to take control of humanity. But if AI becomes more powerful than humans, then AI will control all of us including those who would create them. - FMT

Prof Mohammad Tariqur Rahman is associate dean (continuing education) at the Faculty of Dentistry, Universiti Malaya and an associate member of the university’s Centre for Leadership and Professional Development.

The views expressed are those of the writer and do not necessarily reflect those of MMKtT.