Open letter from top US tech leaders calls for pause in AI development

The recent advances in Artificial Intelligence (AI) showcased by tools like ChatGPT have caused more controversy and debate than any other tech innovation in history. Are we unleashing a monster that could lead to the downfall of humanity, or have we all got too caught up in science fiction movies like Terminator and I, Robot? An open letter signed by 2,000 names including some of the most influential in the US tech sector suggests that it is a question that is at least deserving of serious consideration.

Tesla boss Elon Musk and Steve Wozniak, the co-founder of Apple, are two of the biggest names to appear in the letter, which has called for a six-month moratorium on "the development of systems more powerful than GPT4."

Can we believe anything anymore?

No sooner was the letter made public than many of the signatories either withdrew their support or claimed their signatures had been fabricated – either by AI software or some other means. Others claimed the letter focused on "imagined apocalyptic scenarios" instead of more immediate concerns in the here and now such as AI being programmed with sexual or racial biases.

Setting aside these details, however, the message that comes across with increasing clarity is that software is getting smarter, and that process is getting faster. It is fair to ask – if we do not pause for thought now, will we even be able to later? Or is it already too late?

The letter specifically mentions four considerations:

1. Should we let machines flood our information channels with propaganda and untruth?
2. Should we automate away all the jobs, including the fulfilling ones?
3. Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete, and replace us?
4. Should we risk loss of control of our civilization?

Let's look briefly at each of those points in a little more detail.

Is AI a new source for fake news?

AI's ability to tell a narrative that looks familiar but has no factual basis has already been seen. Deepfakes have been used to create everything from celebrity pornography to political speeches.

In US politics, both ends of the political spectrum have been targeted, with artificially created videos of Joe Biden condemning transgender people and of Donald Trump running from police. Both were fake, and both looked convincingly real.

Handing everything across to AI

The second consideration is one that has nagged away at humans for generations, with every technological innovation. Yet it is one thing creating machines that can dig holes or move boxes or even add up row upon row of numbers so that humans don't have to.

In the past, the argument has been that the machine can do the job faster than the human, and the task is typically repetitive, boring, and often dangerous. So the human can do something more interesting, varied, and rewarding instead – such as overseeing the machine.

With AI, however, it's a little different. The software can compose music that is considered as good as anything a human can come up with. It can write a million short stories in the time it takes Stephen King or Jennifer Egan to write one. But if we even leave storytelling and composition to AI, what is left for us to do?

AI is already better than us at chess, poker, and other things

2008 will be a year enshrined in history. It was the last time a human beat a chess computer. It will never happen again, AI is now far, far better at chess than a human brain. Something similar is happening in poker. With the WSOP, the biggest poker tournament in the US and the world, getting underway in a few short weeks, platforms are having to take extra precautions to make sure online feeder tournaments are only populated by human contenders.

Are we all doomed to obsolescence?

In the second point, above, we talked about AI being more capable than humans in a range of areas. The truth is AI is even better at generating AI than we are. Some people would define that as AI being already in possession of the ability to procreate.

The "Skycorp" premise sounds extreme, but some have argued that all it would take is for AI's goals to diverge from those of humanity. Only last month, Stuart Russell said he asked a Microsoft official if new tools might be pursuing goals of their own. The answer of 'We don't have the faintest idea' will do nothing to set minds at rest.

(Devdiscourse's journalists were not involved in the production of this article. The facts and opinions appearing in the article do not reflect the views of Devdiscourse and Devdiscourse does not claim any responsibility for the same.)