Time Travel: 2014

Why do you say that?

In the previous life, when artificial intelligence was rapidly rising.

Most overseas technology companies have almost unanimously focused on the pursuit of more advanced algorithms, platform framework construction, and commercialization.

"Data annotation" is a field that is neither magnificent nor unique.

Even though data annotation plays a very important role in machine learning, especially supervised learning.

However, the field of data annotation is still dismissed by many overseas technology companies.

Even many overseas giants and some overseas companies specializing in artificial intelligence in previous generations have dismissed data annotation.

Or maybe it’s not disdain, just selective ignorance.

After all, in the eyes of many overseas technology companies, data annotation is a thankless manual job.

Investors don’t know much about data annotation.

They often pay little attention to the field of data annotation.

On the contrary, those technology companies with technology as the core or technology as the core on PPT are more likely to stand out and be favored by investors.

However, the hustle and bustle of artificial intelligence in the previous life is no longer in the limelight.

After taking off the gorgeous coat, we look at the companies engaged in the artificial intelligence industry.

You will find that those overseas companies that once vigorously pursued advanced algorithms, commercialization, and platform framework construction did not necessarily make much money.

(To put it bluntly, most of them are losing money, and they are losing money by burning money.

For example, DeepMind, a company that was a banner in artificial intelligence in its previous life, has basically been burning money after being acquired by Google)

On the contrary, some small overseas companies engaged in data annotation that were not very popular at the beginning have made a lot of money.

There are even some unicorn companies with valuations reaching around US$7 billion.

Although valuation is generally a lot of water.

But as an artificial intelligence-related company, its valuation is about US$7 billion.

After all, DeepMind, which has always been known as the vane of artificial intelligence in its previous life, was valued at less than one billion US dollars when it was acquired by Google.

Under this circumstance, Lin Hui feels that it is not an exaggeration to regard data annotation as a new track in the development of artificial intelligence.

…

By the way, why do all the companies mentioned above refer to overseas companies? Even the so-called "small companies that are not very popular" specifically refer to some overseas companies?

No wonder Lin Hui would single out domestic Internet companies.

Due to some common reasons, domestic Internet companies are basically flowers in the greenhouse.

But the majority of the domestic Internet, except for a few that are relatively capable, is really not very impressive.

Many times when we look at issues from an international perspective, we find that some domestic Internet companies are strange.

It always gives people a strange feeling.

In other words, with high emotional intelligence, domestic Internet companies are generally several versions ahead of the global online understanding.

In many cases, the domestic Internet will take on different forms according to different periods.

Sometimes domestic Internet companies will behave like real estate companies, sometimes they will behave like media companies, sometimes they will behave like car companies, and sometimes they will behave like CX companies.

But it doesn’t look like a technology company.

Lin Hui often simply ignores the monsters that are domestic Internet companies.

If you really want to start a business, go compete with international giants such as IBM and Microsoft.

Competing in the small fish pond of the domestic Internet is actually not challenging.

…

Specific to data annotation.

In the past life, domestic data annotation seemed to have always been a mess.

Because there is no threshold for data annotation, at least it seems there is no threshold.

A college student can basically do ordinary data labeling in less than one day of training.

Such an industry is naturally very demanding.

How many volumes are there?

Lin Hui remembered that in his previous life, he first came into contact with data annotation when he was still studying.

Even crowdsourcing tasks at that time.

You can easily earn 50~70 in just one hour of marking.

Pay day/end, great part-time job.

Lin Hui remembered that there was a time when he was short of money during college and was too embarrassed to ask his family for it.

After half a month of data annotation, I unexpectedly saved some money.

On the eve of Lin Hui's time travel, data annotation with the same intensity could only cost about ten yuan an hour.

It would be great if the salary can be paid monthly (some even in three months), and tax is also deducted.

Rebus is indeed right, if you stand in the wind, pigs will fly.

Many times, even if you can't fly, you can catch up with the bonus period, and you can still have some fun.

Standing on the wind, pigs can indeed fly.

But what happens when the pigs fly?

Can it land smoothly?

The fact is that many pigs that once flew have become nothing more than chicken feathers, no, pig hairs when their bonus period has passed.

The fact is that it only has to do with the Internet.

Regardless of the level, you can't roll well anyway.

But when it comes to data annotation, it’s really complicated.

While wages are increasing in all walks of life in the Internet industry, data show that the wages of employees in this industry have directly shrunk by one-fifth.

It can be said to be simply terrible.

In the past life, it involved such crazy volumes in the field of data annotation.

In many cases, bad money even drives out good money.

Wait until the big companies that own core data realize the importance of data annotation and are ready to give up.

But found that there was no place to even stand.

Even if you have core data.

For data annotation, many times it can only be outsourced.

Many data annotation platforms such as Ferry Public Testing, Goudong Weigong, Ali Crowdsourcing, Gochang Sohuo, etc. are basically products of this type.

It's just so outrageous.

However, this incident also reminded Lin Hui.

If Lin Hui can really become famous in data annotation.

There is no reason why you should not be capable in areas such as interpreting data and data visualization.

That way Lin Hui's tentacles can easily reach other places.

Let’s not talk about these for now, it’s just about forming control over data annotation.

It’s also very impressive.

This almost means that in the future, Lin Hui may completely block the possibility of many companies entering artificial intelligence at the data level.

At the very least, if many companies want to get a share of the artificial intelligence field, they also have to look at Lin Hui's face.

Uh, why do you sound more and more like a villain?

But it doesn't matter, Lin Hui is willing to be kind to others most of the time.

After all, being kind to others is a virtue, but there is a price to pay for being silly and sweet in the volatile Internet environment.

You don’t have to lift the table, but you must have the ability to lift the table.

But these are things for the future.

Although I suddenly realized the economic value of the extremely large-scale text data annotation contained in the past life information and the unique status of the annotated data in the era of artificial intelligence.

Lin Hui's expression didn't show much abnormality.