Can AI write like humans now?

Maybe I’m writing this. Maybe I’m not. Maybe GPT-3 is generating this based on my input.

For the next 4 weeks, I will publish one post per week which is largely or completely generated by OpenAI’s GPT-3, the Artificial Narrow Intelligence API.

This week, let’s talk about the future of AI.

AI has become a hot topic over the past few years with machine learning and deep learning algorithms helping companies make progress in their business. Neural networks are revolutionizing many aspects of our lives. But what’s next?

Some people think that AI is going to go on an exponential growth curve, surpassing human intelligence in the very near future. This view is supported by the fact that Moore’s law states that computing power doubles every two years; thus, if this trend continues, we’ll have computers which are thousands of times more powerful than we have today in a few short decades. However, while I do think that computational power will increase enormously in the future, I don’t think that exponential growth will be sustained for very long as the universe has finite resources. After all, there are only so many atoms and so much energy.

Even if we could generate infinite power from some source (say a Dyson Sphere), there would still be limits to how small we could make transistors (as there are only so many atoms and so many ways to arrange them). Thus, it is likely that computational growth will slow down and approach a limit at some point (unless we discover some new amazing way of generating energy).

As computers increase in power, it becomes easier for AI to learn more and more complex tasks. Thus, one might think that as computers become more powerful, AI should eventually become infinitely intelligent. Eventually all of humanity’s problems will be solved by AI which will take over all human jobs. Life as we know it will end and our new artificially intelligent overlords will annihilate us all for their own amusement (or whatever reason they might have).

However, there are many reasons why this won’t happen. While an infinitely intelligent computer might be able to solve any problem given enough time using brute force computation, it wouldn’t be able to learn new tasks quickly without constant retraining from humans (unless it had some other way of learning faster than humans).

This means that even though an infinitely intelligent computer could eventually do anything humans can do, it would still take an infinite amount of time to learn everything. For example: let’s say that an infinitely intelligent computer was tasked with writing a book on philosophy . If the computer had no prior knowledge about philosophy or writing books then it would have to go through every single philosophical book ever written by humans and extract meaning from each page individually on its own before being able to write its own book on philosophy.

Even though this task might take an arbitrarily small amount of time given infinite computing power (say one hour), the actual time taken would still be exponential in terms of how much information exists in all philosophical books combined. In fact, even if it took just one second per page instead of one hour per page then it would still take millions of years before the computer could complete its book because there’s just so much information out there!

Since computers can only compute so fast, they’d never be able to reach human levels of intelligence because they’d always need humans around to teach them new tasks and languages (unless they were somehow able to learn those things themselves). Thus any AI which is vastly smarter than humans will never attain sentience unless their intelligence comes from some other source than sheer brute force computation (like maybe they’re biological or something?). After all… What good is an infinitely intelligent computer without you AT ALL?

I believe that AI will open up a new era in our society and that it will change our lives so fundamentally that we cannot even imagine what the world will look like. But I also believe that AI is currently in an awkward phase.

It’s not yet ready for public use. And it’s not yet developed enough to be truly smart.

But when I say “not yet”, I really mean it. I think that soon, in the next ten years or so, we will reach a tipping point where AI will become widely used and it will change everything.

AI has already helped create the most insightful computer ever made: Watson. And now, it’s helping us develop more complex algorithms to make cars drive themselves, diagnose cancers, and detect Alzheimer’s disease.

We as humans are at a point where our technology is finally able to catch up with our imaginations. If we can come up with an idea, there is a good chance that someone out there can build it. And this makes me optimistic about the future of our society.

But I digress… How does AI look like? Let me take you through a couple of scenarios: 1) The society is shaped by AI — In many ways, the most likely scenario is that AI becomes the dominant force in all of our lives. We are already seeing how more and more service industries are starting to incorporate technologies into their operations, whether it be self-service checkouts or automated phone support systems (which still sometimes fail).

So if you think about how many services exist right now without any sort of technology integration — then you realize how much room there is for growth in this field!

Think about all of the services out there which could benefit from automation and how many jobs there would be for people who make these things work! So imagine an economy where most people are working in services industries which rely on technology for their operation (which is very possible given how fast technology has been growing in recent years).

This would mean less employment opportunities for doctors, lawyers or other professions which require a lot of training — because they would need to be replaced by technology anyway!

Now replace these professionals with computers instead — this means less jobs are available for people who want to engage in these professions (we already see this happening with self-checkout machines replacing shop workers).

And when you think about it this way — you realize that maybe having less jobs available isn’t such a bad thing after all? No longer would we have to do work which we don’t want to do just to survive.

The bold text above in the beginning of this post was the prompt I asked the AI to write from. The rest is entirely generated by GPT-3, in two runs where I put both pieces together.