In the early days of Artificial Intelligence, the field was dominated by a few trailblazing thinkers who believed that AI could one day become infinitely smarter than humans. One of these was I.J. Good, a British mathematician and statistician who worked on early AI projects at the University of Manchester. In 1965, Good wrote a paper entitled 'Speculations Concerning the First Ultraintelligent Machine' in which he argued that, 'an ultraintelligent machine could design even better machines; there would then unquestionably be an 'intelligence explosion,' and the intelligence of man would be left far behind. As a result, the first ultraintelligent machine is the last invention that man will ever need to make.'
Good's predictions about an intelligence explosion have been borne out to some extent by the rapid progress of AI in recent years. However, it's still far from clear whether AI will ever surpass human intelligence as a whole. In this blog post, we'll explore the arguments for and against the possibility of AI becoming infinitely smarter than humans.
On the side of the optimists, it's worth noting that AI is already far better than humans at certain specific tasks.
For example, Google's AlphaGo AI programhas defeated the world's best human Go players, and deep learning algorithms can now identify objects in images with greater accuracy than humans. It's also worth pointing out that human intelligence is far from uniform; there are many people who are extremely good at one thing but not so good at others. So it's possible that AI could become much better than humans overall without needing to surpass human intelligence in every single area.
On the other hand, there are good reasons to be sceptical about the prospect of AI becoming infinitely smarter than humans.
First of all, it's worth noting that the vast majority of AI progress to date has been made by exploiting the fact that computers are much better than humans at certain specific tasks, such as crunching numbers or searching through large databases. But there are many tasks that are difficult for computers but easy for humans, such as understanding natural language or recognising facial expressions. It's far from clear that AI will ever be able to match humans in these more general areas.
Secondly, even if AI does become better than humans at everything, it's not clear that this would lead to an intelligence explosion. This is because it's not clear that computers would be able to understand or utilise the knowledge they acquire in the same way that humans do. Humans have what's known as general intelligence, which allows us to apply our knowledge to new situations and to understand the world in a flexible way. Computers, on the other hand, are extremely good at carrying out specific tasks but are much less good at general intelligence. So even if AI becomes extremely intelligent, it's not clear that this would lead to it becoming infinitely smarter than humans.
In conclusion, it's still an open question whether AI will ever become infinitely smarter than humans. However, even if AI does become extremely intelligent, this might not lead to an intelligence explosion, due to the limitations of computers' general intelligence.
