There has been a lot of buzz this year around artificial intelligence—more specifically around artificial general intelligence, which is considered human-level intelligence. The enthusiasm for the changes that AI is bringing about in industry, finance, education and healthcare is palpable. It has contributed to efficiency, precision and speed in delivery of services. However, these remarkable developments are being tempered by existential questions on the risks the rapid adoption of AI brings with it. The major areas of concern are threats to jobs, scams generated by deep fakes, misinformation and privacy compromise. And that the influence of AI-based predictions, suggestions and recommendations in our daily lives is proliferating.
AI algorithms cramp us and seek to slot us based on past data and patterns. It stalks and tags us in ways we do not yet fully understand. The potential danger to humanity—the p(doom) factor or the probability of doom—posed by AI has been red-flagged by industry leaders themselves. A super-intelligent machine spawning other super-intelligent machines and ultimately taking control of life and all activities is a probable scenario. There has even been a call to halt research in AI temporarily, until the threat implications are fully understood.
The human species is considered to be at the top of the pyramid of living forms, endowed with a brain comprising about two billion cells. Its significant features are the ability to reason, solve problems, manipulate the environment and use language. Communication, storage and transmission through language has greatly enhanced human intellect. Now, AI driven LLMs or large language models have emerged. Linguistic skills are no longer a human prerogative. LLM models trained to gorge on terrabytes of data have demonstrated the capacity to outperform humans.
So an intelligence that is superior to human intelligence is looming large on the horizon. At this juncture, it would be worthwhile to understand some unique behavioral patterns of other species like plants and animals which helped them thrive, while humans flexed the prowess of their brains.
Animal researchers have documented the agile responses of animals which help them deal with environmental challenges. The network of communication within some species is complex and well organised. By intricate movements of the waggle dance, honeybees direct other members of the hive to the source of nectar. Even the avian, with just a “bird brain”, migrates to faraway sites for breeding, to beat harsh weather and obtain food. The complex organisation of ant nests is possible due to collective efforts and is an architectural marvel. Even trees are known to communicate with each other. Peter Wohlleben, a German forest scientist, has spoken of the “woodwide web”, which is a vast network of fungi called the mycorrhizal network. It helps sharing of nutrients and water between trees. The fungi are also symbiotically benefitted. Such behaviour may be attributed to sensory and biochemical responses. But its effectiveness and dynamism hold lessons for us.
The process of evolution enables the fittest species to survive. Every living form has been through the wringer and has emerged with better adaptive skills. Even non-cellular forms like viruses mutate and find a way to replicate and survive hostile environments. Traditionally, intelligence has been measured by computational, verbal and logical parameters. By those standards, none of the extraordinary abilities displayed by other species may qualify as intelligence. However, the fact that other species have also flourished in the face of an almost brutal human intelligence, is cause for celebration.
Today, there are rapid strides being made in machine learning and neural networks. It is probable that superintelligence may just be a few years away. Sci-fi writers have extensively described various scenarios. Certain aspects like humanoid robots have already materialised. In the future, mankind may be rendered subordinate to a superintelligence.
But there are also challenges that AI has to overcome, chiefly that of availability of data needed for training. AI systems have a gluttonous appetite for data which is a finite resource. Big Tech will need to constantly forage for more data sources.
That brings us to the major advantage human intelligence has over AI. Unlike AI, the human brain does not need a humungous amount of data to learn and understand new concepts. It can handle incomplete information, ambiguities and create explanations. Human intelligence can provide contextual clarity, as it has robust knowledge of the impact of likely outcomes, although AI may outsmart us in computational power. Human control of AI has proved vital in averting disasters in the past. Humanity is thus unlikely to meekly fold up, even as superintelligence spreads its wings.
ALSO READ | Reading our way into the future
I was making a list of the things AI can never do. Topping the list is sleeping and dreaming. AI will never know or appreciate the serendipity of August Kekulé, who discovered the structure of the benzene ring in a dream. AI may never be able to write a poem. It may flounder around with rhymes and line length or manage a few stanzas of ‘found poetry.’ But will it ever match the felt thoughts of an Emily Dickinson or a Subramania Bharati? The theme of the Pondicherry Poetry Festival 2023 happens to be the impact of AI on the niche of writers and poets.
A model like ChatGPT is a storehouse of information. All we need to do apparently is ask it a question. But the moot point is, do all questions require to be answered instantly? A salutary silence could be occasionally allowed to prevail. It would indeed be a comedy of errors when ChatGPT tries to scramble an answer to a kōan.
(ravichandran.geetha@gmail.com)