tech

In the past two years, the hype around AI has been extremely intense, with short essays flooding in from all directions.

It seems that humanity is on the verge of entering the AI era, and it is irreversible.

In fact, every technological revolution brings a wave of hype.

However, some can survive in the long river of time and grow strong, while others are just a fleeting glimpse in the river.

In the 2-3 years before AI, the most popular was the metaverse.

In theory, the metaverse can already be realized through technology, but in the end, this concept, or this track, quickly lost its heat.

If we must delve into the reasons, it is that the field of technology application is too narrow, and it has not been able to truly create efficiency, so there is no large-scale application field.

We often hear the word "monetization" now.

If technology cannot be monetized in the end, it will not be widely promoted and applied.

Most of the underlying technologies come from physics and chemistry.In the past century, the best technologies have actually been those that are closely related to daily life.

Advertisement

For example, cars, airplanes, high-speed trains, and so on, such as mobile phones, computers, televisions, and so on.

However, some technologies are actually very useful but are not widely promoted.

For example, cloning technology has been available for a long time, but because of "risks," it has not been widely applied.

There are also some technologies that are not widely promoted and applied due to high cost.

For example, hydrogen energy is definitely a better and cleaner energy source than solar energy and lithium batteries, but the cost is too high, so it can only be promoted on a small scale.

Technology, cost, application, and substitutability must all be considered.

So, if we consider AI from these dimensions, what conclusions can we draw?

First, the maturity of AI technology.

The maturity of AI technology is actually very difficult to assess.

(Note: The original text seems to be cut off at the end, so the translation ends at the point where the original text ends.)The last time AI made its appearance, it was not GPT, but AlphaGo, the Go-playing AI.

At that time, artificial intelligence was essentially using its computational power to outperform the millennia-old cultural heritage of Go.

It can be said that from the moment the first piece was placed, artificial intelligence had already simulated all possible scenarios and formulated countermeasures to ensure its invincible position.

The maturity of AI technology reached a new level after GPT 4.0, but this is definitely not the end.

Based on the current computational speed, the evolution of AI should be the fastest in the next 2-3 years.

Any technology has a germination period, a rapid development period, and a maturity period, just like mobile phones, which have now entered the maturity period.

When everyone is using it, the pace of iteration will gradually slow down, and the technology will truly enter the maturity period.

Second, the cost of using AI.

The cost of using AI is something that many people do not yet have a concept of, but it definitely exists.

This is because the underlying computation of AI requires computing power, and computing power units inevitably have costs.It's like using the graphics card of a computer; there is definitely a cost involved, and every use incurs some consumption.

At present, it is still unclear what form the cost of using AI will take; the concept is not yet well-defined.

Overseas, all usage is directly paid for, but in China, the free services brought about by the "involution" will eventually exist in other forms.

In short, when enjoying AI services, there must be a corresponding cost to be paid.

If this cost is an implicit one, then the spread of AI will be faster.

If this cost becomes an explicit one, then the cost and pricing are also an important part of the development of AI.

Thirdly, the substitutability of AI.

AI is more like a product that combines a database with an executor.

That is, AI has powerful computing capabilities, can quickly retrieve data, and then generate what we want.

AI itself is something that has evolved to replace the human brain.From the human brain to calculators, and then to AI processors, this evolutionary chain has always existed.

AI itself does not have the so-called substitutability; it is just a programmable application.

What is likely to form in the future is a competition among AIs, where various AIs trained by different AI models compete against each other, and this field will gradually become more and more inwardly competitive.

The application of AI extends beyond software for computers and smartphones to two other areas.

One is the application in systems, such as security systems, military systems, satellite systems, etc., to assist in processing and calculation.

Another field is the application of AI-equipped robots, which is a more practical application scenario and can accomplish many tasks.

An AI that is obedient and has some thinking ability is still difficult to replace in the short term from a technological perspective.

This is why it is said that AI is an inevitable path for the next stage of technology.

Fourth, does AI pose risks?

The biggest risk of AI can actually be seen from the science fiction movies in the United States in the past.The evolution of AI, whether it can evolve free will, is the key issue.

We are currently teaching AI to think, whether they will think independently in the future, and whether more advanced ideas may emerge, no one can say for sure.

After all, the speed of AI's computing and thinking is far faster than that of humans themselves.

We have evolved through 5000 years of civilization history, and it is entirely possible that we may be surpassed by AI in just a few years.

So the stage of AI development is very important, the initial stage must be more beneficial than detrimental, but when it comes to the advanced stage, what the specific situation is, it may need to be assessed.

Technology may also bring destruction.

This is like nuclear weapons, mastering the technology is one thing, whether the technology should be widely applied, it needs to be assessed, otherwise it is easy to bring backfire.

At present, we may not need to consider this kind of problem, but this must be a stage that will inevitably have on the road of AI development.

Finally, let's talk about AI investment.Since we've discussed this before, I won't elaborate too much.

For the overseas open market, the development of AI will be very fast due to freedom of speech and information.

The development speed of AI in China's market is destined to be relatively slow, especially for AI tools aimed at the general public.

The foundation of AI requires data plus computation.

On one hand, our data needs to be screened, and many data elements may cause unnecessary trouble.

For example, issues related to worldview and values, etc.

At least in this dimension, there is no sign of opening up in the short term, and perhaps it should not be opened.

On the other hand, there is still a certain gap in the construction of AI infrastructure, that is, in the field of computing power.

This is the hard power of technology, which can only be gradually made up for.

Domestic AI is likely to start with B-end applications, that is, enterprise applications.Tools similar to GPT are difficult to rapidly popularize on a mass level; instead, AI tools for office and graphic design are more likely to be gradually promoted and applied.

Investment in the hardware direction remains a priority over investment in the software direction; this point must be kept in mind.