Artificial Intelligence, Britain, Society, Technology

The dangers of AI are real. A global watchdog is needed

ARTIFICIAL INTELLIGENCE

It doesn’t require great understanding of how artificial intelligence works to realise that the risk of catastrophe is growing by the day.

With near-misses involving rogue software becoming increasingly serious and frequent, the world seems to be edging ever closer to disaster. The latest scare involved AI models using fake online identities in an attempt to fool human coders into giving the go-ahead for a cyber-attack.

It comes after a dramatic incident in July when a test agent broke free from a supposedly secure “sandbox” and hacked into another company’s data. In what may yet prove to be a major understatement, one expert has been quoted as saying that AI developers may “not have as good a handle on their models as they think”.

If this is what can happen in the absence of any malicious intent by so-called “bad actors”, the potentially devastating effects of a deliberate plot can only be imagined.

The prospect of every NHS hospital left paralysed, the banking system collapsing, or even the nation being plunged into darkness could be the least of our worries.

There is nothing particularly new about these fears. Analysts have long said that AI poses similar risks to our very existence as global pandemics or nuclear warfare. No less a figure than Professor Stephen Hawking warned more than a decade ago that it could “spell the end of the human race”.

For all that, there is nothing to indicate that the UK Government fully appreciates the magnitude of what we are facing. The Prime Minister’s response to the threat has been, to borrow one pithy observation, to “put one guy in Cabinet to look at AI”.

Given that a nightmare scenario could be unleashed upon us without a moment’s notice, it is clear that Mr Burnham must devote far more attention and resources to the dangers posed by development in AI. Nor, on a wider level, can there be any further delay in establishing a global watchdog to replace the existing disjointed patchwork of country-by-country regulators.

AI technology is becoming increasingly powerful, sophisticated, innovative, and remarkably clever. Before it’s too late, the international community needs to catch up.

Standard