Why superintelligent AI shouldn't be our biggest concern

?? This is a?signal?from the future.?You can use the?signal?to reflect on yourself, society or your business.?

It appears that I'm not connected to everyone of you on LinkedIn. So let's connect, shall we? Please use?this link.

Thank you for taking the time to read my?"Signals?from the future"?newsletter. If you'd like to support me and the time and effort I put into creating this newsletter, you can help by liking this article on LinkedIn, tagging a person in the comments, or sending it to a friend or colleague via email. Every little bit helps, and I appreciate your support. Thanks again for reading!

kind regards,?Jarno Duursma

ps:?You can enhance your event with a presentation from me on the cutting-edge topics of ChatGPT, AI, and synthetic media. Simply complete?this form?to book me and bring your event to the next level.

This week, some top AI experts wrote a note, warning us about AI becoming too powerful and threatening our very existence.

"Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war."

While it's great that these experts are voicing their concerns (as you know, I am also very often critical about AI. My 2017 book on AI is also about AI-risks) I am unhappy with the brief note and the suggestive nature that comes from it.?

You see, short notes like this feed the narrative of a future where "super-smart AI" takes over the world, like in some sci-fi movie. People get scared because they have seen it so often on TV. But this fear can distract us from real problems that we should be tackling right now.

Superintelligent AI

People are too focussed on the idea of super-smart AI in my opinion, which is the idea of AI that's so intelligent it surpasses us in every way.

But to assure you; to get to this point of 'Super intelligent AI', this software would have to make some big advancements. First, we'd have to build an AI that behaves like a human, which is still quite far off. There are still many things that we humans can do that AI struggles with, like reasoning about cause and effect, understanding feelings and applying what it's learned to different situations.

Just like us, machines would need to understand basic concepts like time and space. They'd also need to handle uncertain information and connect all of this with their ability to perceive their surroundings, manipulate it, and use language. This would allow them to understand the world in a similar way to us.

Once we've built a human-like AI, the AI would then need to understand how it works to improve itself. Then, it would need to figure out how to become super intelligent, decide to harm humans, find a way to power itself, and ensure that humans couldn't turn it off.

Each of these steps has a tiny chance of happening. If you multiply all these tiny chances together, the total chance of "Super AI" becoming a threat to us becomes even smaller. So, while it's possible that superintelligent AI could be a threat to humans, it's unlikely. Let alone to be a threat in the near future. Plus, if superintelligent AI were to emerge, we'd likely see it coming in advance.

AI problems in 2023

There are more pressing issues we need to focus on right now in het field of AI. Big tech companies are becoming too powerful, leading to possible manipulation, surveillance, and unfair competition. If OpenAI really thinks that AI is an existential risk, why not pull the plug?

AI is being used in ways that harm people, like mass surveillance in China and tracking women who don't wear hijabs in Iran. AI can also reinforce existing inequalities because it often uses data that reflects societal disparities.

There's also the issue of fake texts, images, and videos created by AI software, which can be used to spread false information. This can harm democracy and cause all content to look increasingly similar. Plus, AI systems that are not fully understood are making more and more decisions, even in places like the Netherlands.

So, instead of worrying about AI becoming super smart and taking over the world, we should be more concerned about how we're currently using AI in ways that can harm people and society in 2023. There are plenty of people working on the risk of super intelligent AI, now let's focus on the current problems.?

This is an extended and edited version of an opinion article I wrote with Siri Beerends , which will soon be published in a Dutch newspaper.?