Towards the formation of Artificial general intelligence and what it would look like?

mediumThis post was originally published by Mohith Rai at Medium [AI]

Maybe like this?

A really, really smart friend.

Keep aside the fearful terminator style killer robots for a few seconds. Consider they don’t exist or consider someone found an answer to that. Assume we found a way to figure out how to get an artificial general intelligence working within our lifetime, and it seems to be working. So well that it has transformed the face of the earth, now imagine what that world would look like.

Remember how Tony Stark built his Iron man suit with the help of Jarvis, his AI pal? Tony needn’t have run over engineering drawings or case studies; Tony accomplished some simulations with the holographic display of Jarvis; in fact, Jarvis was doing so well that Tony’s creative input and genius came through.

A companion for you

Some would argue that an Artificial general intelligence might just come up with its own kind of creative input without the need for a human being. I say that it would be more of a separate entity and the design of such a system is within our grasp. We can create how that entity behaves, whether benevolent or malevolent.

I am in the same space as Kurzweil, where he speaks about how AGI could make our lives so phenomenally good that it’s worth taking the risk of complete annihilation. That’s the whole point of being human in some ways, ready to risk an apocalypse for the opportunity to make a paradise on earth. Although it needs to be reiterated that it’s not about changing the world or making the world a better place, some of it is that. Still, more importantly, it’s about substantially increasing the capabilities of human potential and achievement and reach new heights of coming up with awe-inspiring ideas.

It knows everything

Imagine a friend who knows just about everything, ourselves imagine that friend is here to help you achieve your goals. This friend has the cumulative knowledge of the entire human species, and all of that is at your disposal. Do you want to build a fantasy world full of elves and dwarves and need to figure out what kind of language they would speak? The AGI can make a good guess. You need to develop video game characters for your first-person shooter; the AGI can help with that.

Do you have an Idea? Let’s make it work.

Imagine all the times where you’ve had a great idea, and the only way to move forward was implementation. You could have made it work or tested it, and it seems like what is stopping is you are the necessary data to take the next step. How about simulation software? How about engineering expertise to bring your idea to life.

All that and much more possible out of having an AGI focused on designing it. It is said that artificial intelligence is humanity’s last invention. I humbly disagree.

Stark and Jarvis: to build a benevolent AGI will be not done by accident. It needs us to work on a system that works so that the AGI seems to understand what we do and what our goals are. The whole point of great power and great responsibility comes into play here. The AGI is an unimaginably powerful entity that we are dealing with. In theory, we are just projecting the massive hardware and software speed to something that can do what we can. It’s prudent of us to be careful.

Centaur work: After his loss to a machine, which he called a glorified calculator at the time, Gary Kasparov has been working with AI and human teams who seem to be working well together. I had read back then that these AI + human team combo, Centaurs as Kasparov called them, seemed to be doing far better in Chess than both of them exclusively.

This, however, was dashed when deep mind’s AI blew those things out of the water too.

We could be seeing more stuff like this shortly, and the argument is still that AI would not be capable of consciousness.

Data and information: We have unlimited data and information being collected right now, but there’s not much meaning we can derive from it. What we can it’s relatively a small amount as opposed to the amount of data being collected. We really need to develop a way to convert that data and information into wisdom and knowledge. In early 2020 when GPT -3 seemed to be reaching new levels of intelligent AI engineering, many people proclaimed that this was the beginning of AI coming through.

As much as GPT — 3 is revolutionary; it was quickly put aside as people get used to it. Yet, it seems that we are still a few ways off. It doesn’t even seem to be making the news about how smart it is.

Yet the next one, GPT-4, would probably be even much more intelligent than previously assumed.

Perhaps Ray Kurzweil’s predictions are being validated. We may see a new kind of life every 10 years compared to 100 years of transformation. It could be equivalent to going from the Model -T Ford to the Modern cars that we see today in a matter of 10 years, and as we get closer to the Singularity point, that gap would draw much narrower. It’s genuinely going to be a fantastic time to be alive while all this occurs.

We might see new wonders only dreamed of and stranger still than fiction.

As much as we are fearful of what the morrow brings, as long as we have the goals or outcomes that we see for ourselves in the future clear in our minds, we should build a system that would make it work.

Spread the word

This post was originally published by Mohith Rai at Medium [AI]

Related posts