AI is not generative, but synthetic

Getting your Trinity Audio player ready...

This post originally appeared on ZeMing M. Gao’s website, and we republished with permission from the author. Read the full piece here.

Artificial intelligence (AI) is both underestimated and overestimated at the same time. Underestimated for its potential usefulness and potential damages (if used improperly), but overestimated for its real generative intelligent or creativity.

With its efficiency, AI will force people to redefine their jobs.

With its fake ‘creativity’, however, AI will force human being to seriously reflect upon true humanity. If man (male and female) regards himself as just a biological machine, then not only can it be replaced for its functions, but will be superseded in its very existential meaning.

But man is more than a biological machine. It’s up to us to prove it.

What does ChatGPT tell us about AI?

We can learn a lot about AI from the ChatGPT. ChatGPT has taken the world by storm, showing very impressive performance. ChatGPT is not a sudden leapfrog in machine learning, but rather an accumulative result of longtime developments based on GPT (Generative Pre-trained Transformer).

What makes it special is a user interface particularly designed to impress people with an illusion of sentience. The fact that such a design can suddenly create so much impact is itself an indication of misunderstandings about AI.

But I hope more people realize that this type of AI applications (or AI in general for that matter) is fundamentally a knowledge synthesizer, not a knowledge creator.

It is an accelerated way to increase the entropy of the human information system, because it quickly increases the number of the system states (in a thermodynamic sense).

Currently, the attempted rebuttals to ChatGPT all focus on finding evidence that ChatGPT seem to get wrong answers, or at least only get answers that are not as good as an expert opinion.

But these kinds of rebuttals are misguided. It’s quite likely that very soon ChatGPT will seem to get answers as good as even the experts in each field. This would not be surprising at all, not because it validates the claim of AI being real intelligence (or even a superior one for that matter), but simply because it is a natural result of what AI is and does:

It is a human-made program that collects all existing human knowledge with vastly superior efficiency, and further synthesizes it with increasingly clever statistics.

But the fact remains that AI does not really create new knowledge.

Information and knowledge

People tend to confuse information with knowledge. Shannon’s information theory was never about knowledge, but purely information, because information in an abstract sense as measured by entropy does not measure meaning at all, nor does it have anything to do with meaning. It is just about states, possibilities, permutations, etc.

It is about the capacity (e.g., bandwidth in communications), not the content itself.

Information is about existence but knowledge is about order. Information is about possibilities but knowledge has to do with wisdom.

Information is a jungle but knowledge is a path.

Creation and synthesis

Even before AI, human knowledge is 1% creation (through inspiration) and 99% of synthesis (through communications, developments and implementations etc.). Both are necessary, but let’s not confuse the nature of these two.

Human being have been creative (in an exact sense) because of the 1% inspiration, not because of the 99% synthesis. The former is extrinsic (coming from outside of the material), while the latter intrinsic (coming from within the material).

AI is going to vastly accelerate the synthesis part of human knowledge. There is no denying that this is coming and also useful. Given the general performance of GPT, ChatGPT is not surprising at all. It would be even better soon with GPT-4.

But in the most fundamental sense, AI is synthetic not generative.

Human synthesis vs. machine synthesis

The fact that 99% of the human knowledge activities are also synthetic not truly creative causes confusion and an identity crisis.

First of all, if only 1% of human activity is truly creative, does that mean the other 99% of non-generative human knowledge activities can be all replaced by AI?

Absolutely not, and this is the key point that is being widely misunderstood. Even though only 1% of human knowledge activity is truly creative, the entire base and structure of human knowledge is maintained roughly in the following order:

1% creation, 10% interpretation and education, followed by discernment and acceptance by the society as a whole.

In the above knowledge order, not only is the 1% creation important, so are the 10% intelligent interpretation, and the basic human discernment and acceptance.

In other words, although very few of us are truly creative, the fact that true knowledge not only survives but in fact has been intelligently interpreted, discerned, promulgated, applied, and passed down, is very much a testament to humanity. History tells us that although humanity is very much imperfect in doing this, somehow the human society has been amazingly able to retain the key truth without being destroyed by untruth.

This is what is amazing: Despite its fallen nature, the basic humanity has been sufficiently conducive to truth and resistant to the infection of the untruth, so far.

The main reason behind the success of humanity is precisely because a person is not a mere biological machine with a computational component (brain), but a full God-breathing being consisted of spirit, soul and body. See: Why AI Will Never Replace True Humanity.

This is all because human identity goes far beyond what the characteristics of the machine, regardless how smart that machine is.

It means that even though most of human knowledge activities are synthetic not truly generative, human synthesis of knowledge is a very noble thing different from machine synthesis. It is a matter of identity and personality. It is a basis of humanity.

Stop worshiping AI!

Therefore, stop worshiping AI! Doing that is a dehumanizing cult. Learn how to use AI such as ChatGPT as an aggregating and synthesizing tool, but always understand it is just one way to find and present human knowledge to you. It may be more efficient than the traditional way of doing the same, but it is no more than that, and it could even be terribly biased due to the bias of its creators, whether intentional or unintentional.

AI is automation. It can do a whole new kind of nonlinear automation that goes far beyond the traditional linear automation, and can indeed lead to a new Industrial revolution, but it is nevertheless automation, not true creation. Someone said AI should be called IA (Intelligent Automation), and that is a good point.

Will AI change humanity?

AI can never replace humanity, but can AI change humanity?

This is a hard and troubling question to ask. As said above, with its fake ‘creativity’, AI will force human being to seriously reflect upon true humanity. If man does not regard himself as just a biological machine, he will have to prove it.

We will have to prove it on all three identified aspects:

1. Can we still be truly inspired and creative? (The 1% of our knowledge activities)
2. Can we remain truly intelligent in interpreting and educating about true knowledge? (The 10% of our knowledge activities)
3. Can we maintain true humanity by discernment and acceptance of truth as a society? (The rest of our all knowledge activities)

Put another way, the important questions are:

1. Will the 1% inspiration of humanity be sustained or even increased by AI, or will it be reduced by AI?
2. Can the 10% interpretation and education of knowledge continue to be truly intelligent, or will it degenerate into copycats of machine synthesis?
3. Can the society as a whole continue to discern and apply true knowledge facing the increased pace of synthetic knowledge, or will it be overtaken and drowned into the dead machine statistics with a pretense of intelligence?

We can only hope the answer is the former in each above question, for otherwise a terrible end is coming, quickly.

In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.

Watch: New Technologies, New Futures for Nations