Recently, I have been running between China and the United States for a long time. The increasingly serious jetlag (jet lag syndrome) constantly reminds me that I am no longer the young man in his 20s.
This year I have spent an incredible amount of time researching and thinking about AI, to explore how it will change my life, everyone’s life; and how it will change the consulting industry - the most important thing is, I need to do this What kind of preparations.
Coincidentally, it has been exactly 18 months since I last wrote about AI.
(I wrote in "AI is smarter, humans are dumber" on April 15, 2023 : Every 18 months, we become twice as stupid.)
So, have we really become stupid?
In these 18 months, I went through 3 stages.
The first stage is to “cast a wide net”.
To put it simply, use whatever comes out. If GPT is updated, use GPT. If Claude is updated, use Claude. If Perplexity is updated, use Perplexity.
Then try to understand, determine their respective advantages and disadvantages, and personalize them a little bit.
For example, my habit in the past was to open the media, columns, and blogs that I follow or subscribe to every day to read or get news. It probably takes an average of 30 to 45 minutes. I transformed these sources into a structure that automatically obtains, translates and summarizes and implemented it on Notion AI, which ultimately saved my average reading time by 70%. (Let’s not talk about whether reducing reading time is a good thing)
There are many scenarios like this, from reading, to meeting notes, to data analysis.
In short, compared to concepts such as Bitcoin and blockchain, AI is a more pragmatic and key means that allows me to realize that my efficiency is improving every day.
The second stage is "scenario application".
At this stage, I began to think more about how to integrate all the AI tools I have tried into specific application scenarios.
For example, convert a hundred-page consultation PPT with a large amount of information into a readable WeChat tweet; then generate a picture of the key content in the tweet; then automatically convert it into a voice podcast; then summarize and refine it into a video, etc.
Or some bold applications in (other people’s) companies. For example, cost reduction and efficiency improvement in management and decision-making, replacement of some types of work, etc.
By chance, I shared with an entrepreneur how to generate logos, concept maps, illustrations, etc. through different large AI models. After 30 minutes of sharing and letting him try it out by himself, he opened his 20-person design department that night. 16 of them (sorry, my fault) . Three months later, he was very excited to share with me that the work efficiency of the remaining four people after combining AI tools was almost twice that of the previous 20 people.
These scenarios are real and inevitable.
(While I partly blame myself for these people getting fired, in a sense it was really just a matter of time.)
The third stage, starting from the end of September, is also the stage I am in now, called "solve all problems" .
I don’t know if you have heard of it. DeepMind’s original vision was to “solve intelligence and then use it to solve everything else . ”
The first two stages were too fleeting for me.
You will feel that you are exposed to new "products and functions" every day, but often after 1 to 2 months you will no longer feel that you are exposed to new "value".
Fortunately, when I was in a period of confusion, I met a high school senior of mine in New York at the end of September, and it was my conversation with him that inspired me to write this article.
After consulting with him, considering privacy and security issues, I can only share that he is currently responsible for strategic investment in the head AI company (yes, no need to guess, that’s the one) and is the only one on the team. Non-American Chinese.
After knowing his position and what he was responsible for, I had many questions that I couldn't wait to get answered, such as:
How do you (the company you work for) view the future of AI?
How will your own products be iterated and updated?
What kind of investments will you make?
What kind of projects, products, and companies do you think have a future under the big AI trend?
Is the AI capability you understand consistent with what the outside world understands?
After nearly 4 hours of communication, the moment I left the coffee shop where we met, a huge cry began to arise in my heart:
The story of AI changing the world has just begun.
Underlying logic 1: You must believe that AI is omnipotent
The origin of this conclusion comes from when we talked about a very widely circulated article in June called The End of Software . The author of this article, Chris Paik, is a founding partner of Pace Capital, a New York venture capital fund.
Chris compared the impact of AI on the software industry with the impact of the Internet on the content industry:
“Vogue is not being replaced by another fashion media company, but by tens of thousands of influencers; what happens when the software itself no longer needs to make money? We will experience a Cambrian explosion of software, just like Like the explosion of content in the past.”
In this underlying logic, "belief" and "omnipotence" are major premises. (You can always choose not to believe it)
The biggest discussion and question about AI has always been: Can AI really do XXX? Or, is AI not yet capable of XXX?
Users, entrepreneurs, investors, and critics all ask this question.
The answer is simple, just like "Can the Internet really replace traditional media?"
It's just a matter of time.
For example, some people think that AI may be able to write articles now, but drawing PPT is still very rudimentary. This is only a matter of time. AI will soon be able to draw PPTs at the level of consulting companies.
For example, some people think that AI can produce pictures now, but generating videos is still strange. It’s only a matter of time, as the accuracy of AI videos is increasing exponentially every day.
For example, some people think that AI recording and simulated human voices are not similar enough. It’s just a matter of time. Within a year, the AI can even imitate your singing out of tune very accurately (yes, this is one of the projects they are currently considering investing in.)
In short, everything you think AI can’t do, from being unable to do it to being able to do it, should only be a matter of time, not a matter of ability.
So everyone should be worried about:
“If what I’m doing can be done by AI within a foreseeable time frame, what should I do?”
This is a question that deserves to be broken down continuously, and you can even think about it:
In the next 5 years, AI will realize 20% of your professional skills every year. How should you prepare for the future?
Brad Lightcap (COO of OpenAI) said in an interview: "You just have to ask a company if they are excited about a 100x improvement in their model."
The conclusion is straightforward and even cruel:
If you choose not to believe it, you will fall behind 99.99% of humanity.
If you choose to believe but not take action, you will still fall behind 80% of humanity.
Underlying logic 2: The more you give up, the more you get
Yes, what is given up here is "privacy".
"Privacy" has never played such an important role in the field of technology. In any Internet or consumer product of the previous era, privacy concerns were considered more from the perspectives of "security" and "humanities." But in essence, "privacy" never affects the functionality and practicality of a product, or in other words, it does not affect the functionality and practicality of a product to a great extent.
But AI products are different. Regardless of any product, this will be a formula in which the more you give up, the "customization" you feel will rise linearly, without exception.
This has become an almost unsolvable proposition: the prerequisite for “understanding you” is “knowing you well enough”.
Therefore, there is an interesting saying circulating in the Silicon Valley AI circle that the best AI agent application scenarios and commercial implementation are likely to start in China. (Do you ever feel like "you are so dirty when you curse"?)
For example, we talked about Carbon Robotics, which was very popular in the past few months. Its core product, LaserWeeder, is an automatic laser weeder that uses AI technology and a high-precision laser system to accurately identify and remove weeds in the field.
The premise is that farmers give up information about their crop types and their locations.
In fact, when I wrote this, I felt that this might not even be a good example, because location information may not have a strong relationship with personal safety, behavior and habits, so try changing the subject. What if it was your love experience? What if it was your daily routine? What if it was the source of all your information intake?
If you feed your entire life to an AI assistant that can become "omnipotent" over time, you will get an assistant that understands you best and is the most efficient.
This kind of derivation will make people feel very interesting (neutral, not complimentary) .
In his latest article "Writing and Not Writing", Paul Graham (co-founder of Y Combinato) wrote that due to the impact of AI writing, in a few decades there will be very few people who can write.
This is just like before industrialization, most people were strong because of labor, but now if you want to become strong, you have to take the initiative to exercise and keep fit. The same is true for writing. There will still be smart people, but only those who choose to think.
This loop on "privacy" will form rounds of discussions in the next 10 years. People who insist on privacy will be very rare, causing you to have to spend extra energy to get what you should have.
Underlying logic 3
(I'm not going to share it)
Unfortunately, the third underlying logic is the only part I chose to leave out in this article.
I believe that the first two underlying logics have allowed enough people to start thinking about their future choices, and I want to leave part of the insights for a time when "I can let the bullets fly for a while."
As a teaser, what I can share is that this is the logic regarding "minimum professional age".
(You can search for the news that Chat Nio, a 15-year-old high school student, was acquired for one million yuan.)
In the foreseeable future, “elements” that the past education system could not carry are being realized one by one. Young people in the future will have fundamental concepts about education and career that will change, just as the career of today's younger generation of children actually starts in school.
So the question you can think about is: Is basic education (traditional form of education) still needed in the future?
Finally, let’s get back to the “should be anxious” thing.
I deeply believe that this is a real anxiety that needs to be created, because the information gap is too great for 99% of people, leading to a high possibility that we will experience submersion without realizing it in the torrent of the times. .
This change is not simply a change from paper to electronics, but a structural change in the economy, affecting everyone's lifestyle, career choices and professional skills.
I would like to end with a project that can be shared within the scope of confidentiality. This project is also the first project that the senior invested in during his employment, called The Perfect Teacher (pseudonym) :
Imagine if all your children’s electronic devices, whether at school, at home, or in cram school, could be “omnipotently” monitored without any conditions. The Perfect Teacher will analyze the child's overall learning efficiency and when there is a gap, give suggestions and adjust the learning mode, and provide feedback on learning results and evaluation reports to teachers and parents at any time.
As a parent, how much are you willing to pay for this? Would you be willing to sacrifice your child's privacy because of this?