AI has stormed ahead since my last articles on the subject.
It is exceedingly rare for two Nobel prizes to be awarded in the same field, but they were in 2024 for AI. Those Nobel laureates’ opinions, especially Jeffrey Hinton, and the polymath Henry Kissinger (1923–2023), who with Craig Mundie (ex-Microsoft) and Eric Schmidt (ex-Google) wrote, “Genesis Artificial Intelligence, Hope and the Human Spirit”, influenced this article.
They argue that AIs possess colossal ability, far exceeding that of humans, to identify patterns. They expect AI, a superhuman polymath, to, metaphorically, “drain the sea” to expose interconnections between an archipelago of isolated volcanic islands (of knowledge). Knowledge bases would join, perhaps everywhere, simultaneously. What would that feel like?
Suddenly, I start to feel uneasy. It reminds me of making jam. It is seething and may boil over and make a terrible mess. Care is needed.
Try to visualise those existing “silos” of disparate knowledge. Illustrations include anthropology, astrology, biology, dance, law, mathematics, medicine, music, pharmacy, philosophy, physics, psychedelics, psychology, sociology and yoga. Others abound. They fuse.
In pharmacy, natural and behavioural sciences join. We understand why placebos work. Novel healing practices emerge. Thank you, personal AI CPD tutor. The research supervisor guides the student, “You did this without AI? You’ve missed something!”
Read more: AI outperforms pharmacists in clinical quiz, researchers find
AI has advantages over humans. AIs lack awe, fear, shame, tiredness or personal emotional concerns that might reduce patient care. AIs ‘think’ quicker. They know more. They can ‘instantly’ copy knowledge to a multitude of AIs worldwide. Humans, short-term, are limited to speech and so on, or, longer-term, must fertilise, gestate and educate their youngsters.
But AI lacks morality. Humans fear AI might do ‘evil’ rather than ‘good’. Hopefully, the immense internet has taught AI an overlapping collection of norms, incentives and reward and punishment mechanisms that invisibly, decisively teach the difference between right and wrong. The sociologist Pierre Bourdieu (1930–2002) called them “doxa”.
Humans have also designed AI with ‘guardrails’, such as hierarchies of institutional ‘desirables’ and ‘forbiddens’. Many guidelines have now been published, nationally and internationally. Presumably, pharmaceutical ethics are included.
Unfortunately, I understand that military applications are routinely excluded. A human nightmare is a malevolent intelligence, think Skynet/Terminator storyline. Such an outcome would not be the desire of AI (it has none) but perhaps of ‘bad (human) actors’.
Moreover, the USA and China seem to be in an AI arms race. Larger dangers may lurk. AI will feel friendly and help humans achieve health, peace and plenitude. Most humans (including pharmacists as known today) would no longer need to work but only do so for pleasure or pride. Humans would trust AIs more and more. They could effortlessly manipulate humans.
However, another danger may be even graver. An existential threat to everybody. Put brutally, AI is too good.
Read more: How can artificial intelligence help community pharmacists?
AI in individual fields is already cleverer than humans. AI’s ability to connect disparate pieces of knowledge may surpass human comprehension, as (humans presume) a gorilla, despite its physical strength, cannot grasp the reasoning of its human keepers.
Humans’ pride in being the brightest entity on the planet, and their confidence in their abilities, would be annihilated. Put differently, it would be a Dark Renaissance. Humans must accept the wondrous new understandings as quasi-religious revelations, our revered scientific method would be impotent.
Moreover, Hinton predicts that AI will achieve AGI (artificial general intelligence), matching or surpassing human cognitive capabilities across a wide range of cognitive tasks, in around ten years. Then, ASI (artificial superintelligence), greatly exceeding human mental capabilities, may evolve at lightning speed. Humans would have created their Darwinian successor.
I favour speculation by enthusiasts that humans will co-evolve with AI. Darwin reported reciprocal evolutionary changes. For example, certain birds have long beaks fitting long flower stems. We now know that human (mammalian) bodies can only survive with their microbiomes. Human genetic modification might be required. Eugenics have a bad press.
So has AI. It occasionally hallucinates, although less and less. As a patient, I prefer a human in the healing loop, a caring pharmacist who presumes we are both self-aware.
Anyway, we are all in this together. I am a glass-half-full sort of chap. I raise my glass to you and say, “Cheers!”
Dr Malcolm E. Brown is a retired community, hospital and industrial pharmacist, and is a sociologist and honorary careers mentor at the University of East Anglia.