Artificial Intelligence can reach the IAG level and this is dangerous!

Although AI has proven to be quite intelligent to date, it is still not as intelligent as a human being. In fact, we are far from that point, but there are people who believe it is closer than we think. According to Google’s head of AI, there is a 50/50 chance that AI can reach the IAG (Artificial General Intelligence) level, and this could happen sooner than we think.

Artificial Intelligence can reach the IAG level and this is dangerous!

There is a lot of fear around AI these days. However, if you are afraid of AI, then you don’t want to know what AGI is. Basically, AGI is an AI that is as intelligent as a human. Modern AI chatbots like ChatGPT are great at regurgitating the information presented to them. Even if it is generating a story it still does not represent true human intelligence.

Google’s AI chief still believes there’s a 50/50 chance of achieving AGI in the near future

In 2011, Google’s current head of AI and co-founder of DeepMind, Shane Legg, publicly stated that there is a 50% chance that AI will progress to AGI in the next two decades. Now, more than a decade later, he hasn’t moved away from that idea.

It was in an interview that he expressed his ideas about IAG and when we could expect to reach that point in development. Although he stated his position in 2011, his interest in AI goes back another decade, when he read 1999’s The Age of Spiritual Machines. He pointed out two main points in the book that really resonated with him. “One is that computing power would grow exponentially for at least a few decades. And that the amount of data in the world would grow exponentially over a few years.”

The book was not wrong in this regard, as the amount of data produced every day is increasing with each passing day.

Difficulties in carrying out the IAG

The author predicted that IAG would be available later this decade, but later said that this might not be the case. One of the obstacles is the fact that it is difficult to define what human intelligence is. It’s a very vague concept. “We will never have a complete set of everything people can do.”

However, he also said that there would be no single test to see if an AI is smart enough.

Another point mentioned by Legg was the fact that we are not able to increase the scale of AI models to take AI into IAG territory. Right now, the AI ​​models that power today’s chatbots are absolutely gigantic. They’re unimaginably big, but they’re not big enough to create true AGI, according to Legg.

So how do we achieve this technology?

At this moment, Legg believes that we are on the right path to creating an IAG. Now it’s a question of knowing how to train our models. Legg believes that companies should “start training models now with the scale of data that is beyond what a human being can experience in a lifetime”. It’s an ambitious request, but he believes it’s achievable.

Right now, as we wait to see if humanity will create the first AGI, companies are still working to make their AI smarter.