Roadmap shows the trend of AI “replacing” human occupations

I saw an interesting picture yesterday, it was a "level map of AI replacing human paths".

Picture

As shown in the picture, the game is divided into six different levels, from E1 to E8.

We can observe from the figure that artificial intelligence (AI) is replacing human applications in different fields. The application field path of artificial intelligence is determined by its fault tolerance rate.

In short, the error tolerance rate here refers to the cost of trial and error.

AI will gradually replace industries with higher to lower fault tolerance rates, and gradually "replace" human occupations.

In the past, we often thought that creative work relied on human thinking and was not easily replaced. However, with the development of artificial intelligence, this view does not seem to be entirely correct.

Creative jobs often have no fixed answers, and those jobs without clear answers are often more likely to be replaced by artificial intelligence.

Why is this so? This may involve the issue of fault tolerance. Often, the more jobs there are without standard answers, the customers may be less picky about the work. If the customers are not so picky, the cost will be relatively low.

It is difficult for AI to replace jobs that have clear and strict standards, because these standards are unified and customers are clear about them. Regardless of whether the performance is good or bad, they will be very picky because the cost and price are very high, so they dare not easily let AI intelligence replace them.

The same applies to autonomous driving. For autonomous driving technology, we have strict standards. What conditions and standards must be met before it can be put on the road. Otherwise, once a liability accident occurs, life safety will be involved, and even How should responsibilities be divided? This may not even be clear.

Autonomous driving has always been far away because the error rate cannot reach the expected value (only the same order of magnitude as humans). The subject of responsibility is a very serious issue, so the fully automatic L4 error rate must be reduced to a low enough level to be able to cope with sky-high compensation.

The same is true for accountants. No matter how well written the GPT results are, the accountant still needs to sign and take responsibility as the responsible person. If the AI ​​error rate is on the order of 1%, it will still require human inspection and verification. In fact, there is no significant improvement compared with current tools in the accounting field, and it is even inferior to existing tools in terms of accuracy (existing tools can also automatically import files , there are also many checking functions from various angles).

So from high to low error rate, it is easy to see the timeline of AI “replacing” professions.

Some time ago, I was renovating a house. The bricklayer was a relatively young man in his 30s. He said that it was enough for me to do two jobs a month. Two jobs would basically make over ten thousand. The life was quite comfortable.

At that time, I was chatting with the bricklayer about this matter. I said, your job is pretty good, you can make a lot of money, and it won’t be easily replaced by AI.

After all, tiling is a job with very strict standards, but it is difficult to be quantified by AI. If you look at current jobs, the jobs of people sitting in offices are easily replaced.

The bricklayer master said: There are very few bricklayers now. Except for some old masters, young people are even less willing to do it.

I said: Then your value may become higher and higher in the future, and things are rare and valuable. What’s more, it’s difficult for AI to replace your job.

While chatting with the master bricklayer, I also said that in fact, some jobs cannot be replaced by AI, but the cost of replacement must be considered.

So, in addition to fault tolerance, replacement costs also need to be considered.

It is the work cost of AI work > labor cost, and AI will not replace this type of work.

Why do I have this idea? I chatted with the master bricklayer, and I said that when you lay tiles, you need to ensure that the tiles you put out are relatively flat. To keep them flat, AI may not be unable to do it, but The cost may be relatively high; for example, different houses have different sizes, and the owners have different requirements for laying ceramic tiles. At this time, even if there is a machine, the designer may need to obtain a large amount of measurement data before feeding For machines, the cost may be quite high.

In fact, there are many details involved in laying ceramic tiles. If every detail was extracted and assigned to machines, the cost would be extremely high.

So, if the details involved in the work you are engaged in are very complicated, and each detail has strict standards, it will be difficult for this type of work to be replaced by AI. Not only is the fault tolerance rate low, but it is also more likely to be done by AI. The cost is higher than labor cost.

What do you think of this?

The above is the detailed content of Roadmap shows the trend of AI “replacing” human occupations. For more information, please follow other related articles on the PHP Chinese website!