When Could AI Overtake Human Performance?

Juan Umbarila
The question no longer seems to be whether it can do it, but how soon, and how will it happen.
The path to an Artificial General Intelligence can make us reflect on how we asses human intelligence and performance. Photo: Andrea de Santis | Unsplash

It is a fact that Artificial Intelligence is going to create a major disruption in human history as far as technological breakthroughs go. If we take a look at AI’s exponential development, improvement, and adoption, and how it is already being used, the short answer is that it is already overtaking human performance in several instances.

A 2022 survey of AI experts predicted that there is a 50% chance that human performance will be surpassed by AI in 2059. That is eight years sooner than what was predicted when the survey was last made in 2016.

By “outperforming,” the survey means a level of high machine intelligence observed “when unaided machines can accomplish every task better and more cheaply than human workers.”

But the long answer demands that we take into account that AI overtaking human performance depends on variables like the area of knowledge/application and the level of specialization of the task. In other words, the process will be asymmetrical, and dependent on the real-life conditions of its use.

This also demands that we define what we mean by human performance, and whether its surpassing by AI means an existential threat to humanity, or possibly a way to enhance it even further.

Happening In A Not So Distant Future? Or Not Quite There Yet

AI intelligence advancement depends largely on technological improvements and computing power. Photo: Julien Tromeur | Unsplash

One way to measure performance is by the job market, currently one of the most prominent fears regarding AI. According to Forbes, three industries, in terms of human jobs, are under immediate threat by AI implementation: finance and banking, media and marketing, and legal services; while manufacturing, agriculture, and healthcare will be the least affected in the short to medium term.

Goldman Sachs estimates that around 300 million full-time jobs could be lost to automation.

Read: Prominent AI Researcher Quits Google & Regrets His Work.

But these are all specific tasks that AI can be programmed to do by being trained on (relatively) limited datasets. If we mean overtaking human performance in a more comprehensive, general sense, AI technology needs to advance to a stage regarded as an Artificial General Intelligence (AGI) which is not yet entirely foreseeable.

In general terms, AGI means a human-like intelligence in a cognitive sense, that may or may not include a conscious state of self-awareness. Depending on whom you ask, this intelligence could be centered on achieving certain goals, or on general adaptability, intuition, and survival.

In any case, achieving an AGI is not only contingent on the amount of data fed to the machine, but to a level of computing power that is not yet available, and that may only be achieved by the future development of quantum computers.

According to what Dana Rezazadegan, an expert in AI and Data Science, says in an interview with Science Alert: “With quantum-enhanced AI, we’ll be able to feed AI models multiple massive datasets that are comparable to humans’ natural multi-modal data collection achieved through interacting with the world.”

Is This Intrinsically A Bad Thing?

IBM’s Deep Blue computer defeated human chess world champion in 1997. Photo: Erik Pitti | Flickr

There are precedents of machine intelligence surpassing human abilities in certain tasks that have not led to disadvantageous conditions for people. One example is the chess world champion Garry Kasparov being defeated by IBM’s Deep Blue computer in 1997.

This event did not fundamentally change the human relationship with chess in a bad sense; in fact, it can be argued that it led to human-machine collaborations that have furthered human abilities in the game.

And this applies to a certain extent to more traditional jobs as well. According to a research paper on whether AI was to replace human radiologists, the results showed that human job displacement was a mirage for this area, and that human radiologists were enhanced by AI rather than replaced by it:

“‘Will AI replace radiologists?’ is the wrong question. The right answer is: Radiologists who use AI will replace radiologists who don’t,” said its author.

Indeed, the other side of the debate sees AI surpassing human performance not as a threat to human existence or even our jobs, but as an opportunity for human-machine collaboration to improve people’s lives.

Sam Altman, CEO of Open AI, considers that “If AGI is successfully created, this technology could help us elevate humanity by increasing abundance, turbocharging the global economy, and aiding in the discovery of new scientific knowledge that changes the limits of possibility.”

Once again, it largely depends on how humans implement the technology. After all, machines do as they are programmed, and a founding principle of AI is to serve humanity, not replace it.


Leave a Reply

Your email address will not be published. Required fields are marked *

Related Posts