Advancements in Artificial Intelligence — the capability of machines to make informed decisions and perform tasks usually reserved for humans — are moving at a rapid rate, and it’s threatening workers from truck drivers to surgeons, according to a new study from the University of Oxford.
Spearheaded by Katja Grace of the Future of Humanity Institute at Oxford, the report surveyed more than 350 AI experts on how long it’ll take machines to master certain jobs, from remedial to advanced. The results are jarring.
Within the next decade, experts predict machines will outperform humans when it comes to translating languages, writing a quality high school essay, and driving trucks. AI proficiency in sales and retail is expected by the early 2030s.
The essay data is especially intriguing. It indicates AI is on the cusp of routinely passing the Turing Test — a baseline trial where humans are unable to distinguish between interacting with machines and another person.
High-level jobs are not far behind, either The study anticipates machines will be superior to humans in writing best-selling books and performing surgery by 2053.
Altogether, 50 percent of the researchers surveyed believe we’ll achieve “High-level machine intelligence,” or HLMI — where unaided machines can accomplish every task better and more cheaply than human workers — within the next 45 years. Of course, experts can often be wrong, but even the high-end of these predictions anticipates automation of all jobs within the next 120 years.
Reaching HLMI will be a critical fork in the road moment; important scientific and technological breakthroughs could be offset by “catastrophic risks,” as the study puts it.
On the one hand, this leap forward can lead to a cure for AIDS or put an end to food shortages. “The power of AI technology is it can solve problems that scale to the whole planet, said Mike Schroepfer, Facebook’s Chief Technology Officer.
On the other hand, 5 percent of experts indicated it can lead to an “extremely bad” outcome — like human extinction — if the AI is not harnessed properly. 48 percent of respondents said research on minimizing the risks of AI should be prioritized.
Neuroscientist Sam Harris expanded on this threat in a TED talk last year. “It’s not that our machines will become spontaneously malevolent,” said Harris. “The concern is really that we’ll build machines that are so much more competent than we are, that the slightest divergence between their goals and our own could destroy us.”
Still, even with machines working in perfect harmony with their human overlords, the study’s predictions point to a looming issue: how mass automation will affect the human psyche.
A 2014 Gallup poll found “unemployed Americans are more than twice as likely as those with full-time jobs to say they currently have or are being treated for depression–12.4 percent vs. 5.6 percent, respectively.” The figure increased to nearly 20 percent for the long-term unemployed.
In other words, as workers are replaced by machines, it’ll be imperative to untether self-worth from employment.
To read the full “When Will AI Exceed Human Performance? Evidence from AI Experts” study, click here.