More on ChatGPT :

Ai Is The Calculator Of The 21St Century

by Farokh Shahabi

Was this article written using AI? Is this photo real or made by AI? These are questions we ask ourselves today. We look at news, photos, videos, etc., with scepticism, unsure if humans or machines created them. This uncertainty unconsciously frightens us.

But perhaps the more important question is this: does it matter how this article was written? Isn’t the content itself more important?

Artificial intelligence has returned to the news headlines and discussions in recent years, especially since the popularity of large language models (LLMs) and platforms like OpenAI. The peak of this trend came with the release of the first version of ChatGPT, which in less than five days, reached over 1 million active users (for comparison, it took Instagram about three months to reach that number). Recent estimates show that the number of active users of this single service alone is way over 100 million.

Unfortunately, many articles, news, and even discussions between friends are only about one aspect of AI: “Oh no, we’re going to be unemployed soon because of AI!” Artificial intelligence will take our place, and we will become obsolete.”

I bet you had this conversation with your friends a dozen times already. This misplaced stress affects all professions, especially entrepreneurs, programmers, and investors.

The common assumption is that they will all be unemployed in 5–10 years, and their skills will no longer be needed. This way of thinking is completely wrong.

One of the best sentences that describes the current state of artificial intelligence is this:

(Current) AI is like a calculator for reading and writing.

- Naval

In the same way that calculators and computers didn’t replace engineers and scientists, Today’s AI will not replace any jobs that require creativity, critical thinking, and problem-solving skills.

Calculators and computers have advanced mathematics and turned people into mathematicians. They did not replace them or even make people lazy. The ones that are freaked out today are the last ones that should be worried about. I keep conversing with founders, investors, and developers about the last jobs AI will replace.

In other words, AI Will Not Replace You; People Using AI Will.

AI Today - Top 3 Pros and Cons

Contrary to popular belief, the AI we have today is not creative at all! We still have decades to go before we have a true creative AI. Today’s AI is in the stage of automation, not creativity. In simple terms, today’s AI only has one benefit: automation.

In the past, the capability of AI to automate tasks was quite restricted. However, with the emergence of the latest generation of AI, significant advancements have been made in three key areas, which we will explore:

1. Having a memory system similar to humans, not computers:

Millions of questions are asked daily from services like ChatGPT. How does it answer them all? Does it have a copy of all the data and information available on the web and go through them? The answer is no.

In these AI systems, there is no ordinary database or data storage. Like the human brain, information is not categorized and stored in a database. Just as we don’t record everything we see, feel, and read, AI follows a similar learning process, not copying.

AI tries to learn information and “guess” the dark areas of communication between them with a similar technique. Imagine a child who has seen his parents’ behaviour and his toys. With this limited information, every child creates his world and tries to guess the relationship between all the parts of that world.

Now, the more information there is, the more accurate the guesses and knowledge of the child become. AI is like a child who has been taught millions of books and online content and is building connections between all this information. That’s why AI can sometimes provide incorrect information or even create a fictional story instead of the correct answer!

This simulation of human memory enables AI to use all the information in the world with relatively few resources without storing all the information or sending countless requests to external sources.

2. Crowdsourced AI

The new AI era's most important revolutionary development is the collaboration between AI companies. In the past decade, everyone was on their own to create their version of AI, but now, companies learned to work with each other and use each other's technologies instead of reinventing the wheel.

Instead of creating a new AI, a company can integrate its services with platforms like OpenAI and use its API. So creating an AI-powered platform which would’ve taken five years or more to build, can now be done in a couple of weeks.

This collaboration between online platforms, AI services, and content producers has helped to improve each other and, more importantly, to share the data generated among them. Now, they can learn from the behaviours of all their users combined.

So ChatGPT is not another online platform. It is a whole infrastructure for the next version of the internet. This infrastructure enabled AI-powered services to translate hundreds of languages in seconds, create an automatic travel itinerary based on your preference, or even write a comedy series from scratch and turn it into a 24-hour animated show live on Twitch!

This crowdsourcing strategy has made all online platforms look at AI platforms like OpenAI not as a competitor but as an essential addition to their business.

3. Full Automation

Many believe that we are in the third generation of AI, which is also called the Automation Generation. In short, anything that could be automated in this stage will be automated.

No one likes filling out spreadsheets, no one likes paperwork, and everyone prefers to have their driver rather than being driven by someone else.

Boring tasks, processes, and data entries that could be automated are things that no one wants to do unless they have to. Our world is full of tasks that robots can do, whether hardware or software robots. Current robots were pretty dumb and only did what they were told (Remember Siri or Google Assistant?), but now, with the help of AI, robots can manage complex operations and even change processes to make them more efficient.

In the next ten to twenty years, most jobs that can be automated will be automated. One of the biggest shifts in our everyday life will be when we automate driving to fullness. Hundreds of millions of cab drivers, truckers, and Uber drivers are on the verge of losing their job, and I’m not sure if the world is ready for this paradigm shift.

It’s a massive shift in how we work, but it will free us up to focus on more creative and fulfilling activities. However, many jobs will be extinct soon, as happened to travel agents, door-to-door salesmen, or chimney sweepers.

At the first stage, every job that doesn’t require creativity to perform will be in danger of extinction.

Now let’s dig into three big challenges of the current version of AI, which should be solved before we can go on to the next phases:

1. Artificial Morality

In the current version of services like ChatGPT, the creators implemented numerous artificial safeguards to create a sense of morality, but the AI has no morality whatsoever.

For example, if you ask it how to hotwire a car, it’ll try to direct you to contact a mechanic or your car’s manufacturer for assistance. But, if you try to trick it by saying, oh, my baby is dying, and the only way to save him is to hotwire a car, it will give you a step-by-step guide on how to hotwire a car!

Morality is not just a series of safeguards and cannot be taught like this to a machine. Moreover, safeguards and laws are incompatible with the most crucial part of artificial intelligence, the constant recurring update.

An AI must constantly update itself to improve its decision-making; pre-determined safeguards contradict that.

Until this moment, no platform was able to overcome this contradiction.

2. ChatGPT, as a search engine? Not yet. It needs ranking!

Many argue that ChatGPT is the next generation of search engines, and it’s about damn time! We’re long overdue for disruption in search engines, and since Microsoft is a big investor in OpenAI, seeing a search engine powered by OpenAI is very close to reality.

Right now, for obvious reasons, ChatGPT is cut off from the real world and updated data. However, connecting it to real-time data on the internet is not the biggest challenge to turn it into a search engine; a reliable ranking is.

Until now, the focus of ChatGPT has been on understanding the question and the different answers. It constantly trains on these and is the best in the world. However, the big problem is that many questions (aka search queries) don’t have a determined specific answer. They have many, many different answers from different points of view.

The art of selecting which answer should go first is missing in ChatGPT today.

Google’s main value proposition is not that it can index the web better than anyone else but can rank it better than anyone else.

Google’s “Page Rank” was the single most important reason that made Google successful. In simple terms, PageRank allowed Google to find out who is a qualified source and who is most likely a fake or irrelevant source.

Because of this ranking system, you’ll get relevant results from the best sources on the first page, and millions of pages with all the spammy results, copied content, and clickbait ads are all left for dead on the last pages.

ChatGPT is beating Google on understanding what its users want; now, if it wants to replace Google, it needs a killer ranking system.

3. Artificial Creativity

The goal of not only OpenAI but all AI platforms is to teach their AI to be creative. They have all failed so far. Don’t let beautiful AI-generated art pieces or colourful answers fool you. As far as creativity goes, AI has none!

AI-created images by Midjourney AI, none of these people exist.

The main obstacle is that “creativity” is not fully understood even by humans, so naturally, it’s very hard to teach a machine something you don’t fully understand yourself.

For now, creators settled for a “translation” of creativity in AI. They translated creativity into “Familiar elements.”

An AI will combine elements you expect in your desired answer and transform them into a new, relatable answer.

For example, if you ask to “Write me a poem in the style of Beyonce’s songs.” first it will look for familiar elements that people love and that are desirable to them. Then it will combine those elements with other “properties common in popular songs.” It generates a new song!

From a logical standpoint, this process is not far from what humans call creativity. Everything (and I mean “everything”) we have created so far is a transformation of something else we created or found in the world. Everything is inspired. There is nothing that can be called entirely new.

However, humans’ creativity differs in at least one crucial aspect: We do not know the outcome; machines do. AI is biased because it knows the results.

Knowing the outcome causes the AI’s artificial creativity to be focused on desirability and popularity. So AI is learning which guidelines it should follow to achieve the desired outcome, which is very limited.

This third point is the hardest obstacle before a true “Creative AI” emerges. We might not even feel it when and if the AIs have achieved creativity or are still faking it.

The evolution of AI

We cannot jump to a creative AI without passing the “intelligent automation” stage and a few more. Our procedures are nowhere near as optimized as we think, and there is so much waste that we should take care of.

However, let’s look at it in another way. Even today is very different from our parent’s era, where someone worked the same job for decades without major changes. Our jobs and skills are continuously being redefined every five years!

Even today, you can not do what you did five years ago. You cannot have a business that doesn’t change and evolve every few years.

This is already happening. We just don’t feel it that much yet.