Why the odds are against AI doomers

In 1968, Paul Ehrlich warned in his book, The Population Bomb, that the battle to feed humanity had been lost and millions would starve regardless of the measures taken. However, the Green Revolution achieved significant progress between 1960 and 2005, with major cereal crop yields more than doubling and real food prices falling by over 50%.

Ehrlich, who died in March at 93, now feels the same concerns, this time regarding artificial intelligence (AI). In a 2024 Pew Research Center survey, 51% of Americans were more concerned than excited about AI, while only 15% of 1,013 AI experts shared this sentiment. Over half of the experts expected AI to have a positive impact on the United States within the next 20 years, compared to just 17% of the public.

Malaysians also show mixed feelings, with 77% excited about AI, but 55% nervous. Despite these concerns, researchers building AI technology are optimistic. In a 2023 survey, 2,778 AI researchers estimated the median chance of an extremely bad outcome, such as human extinction, at 5%, while 68.3% believed good outcomes were more likely.

Examples of AI's positive impact include Google DeepMind's AlphaFold mapping over 200 million protein shapes, earning creators a Nobel Prize; AI-assisted breast cancer screening catching 29% more cancers with 44% less radiologist work; and AI at the University of California San Diego reducing hospital sepsis deaths by 17%. The benefits extend to customer support, where AI assistants improved productivity by 14% and for novice workers by 34%.

In Malaysia, AI is already being used to screen diabetics for retinopathy and assist in chest X-rays, with government clinics piloting AI-assisted imaging to detect lung disease earlier at a one-time software cost of about RM10,000 per clinic. The writer cautions against fear while advocating for preparedness, emphasizing the importance of building safeguards and retraining workers as AI continues to evolve.