Focus
1. The iPhone makes it easy for non-tech-savvy people to get started, and has developed a whole set of app store models, ChatGPT in this regard, very similar to Apple's iPhone;
2. The ChatGPT pandemic, not only because it brings a lot of practical features, but also because it is very human-like, which leads people to prefer technology when it has obvious human qualities;
3. Anthropomorphic artificial intelligence will hinder us from entering other artificial intelligence paths, and the development direction of artificial intelligence is to enhance various human capabilities, rather than having to imitate humans, nor is it recommended that industries pile up large models;
4. Without some kind of global regulation, the AI race will not stop, because it is based on business logic, Musk and others signed an open letter hoping to stop development for six months, basically wishful thinking;
5. The longer-term and more worrying issue is the application of AI to weapons to automate weapons for mass destruction.
overview
In November 2022, OpenAI officially launched the generative pre-training model ChatGPT-3.5, which gained 1 million users in just 5 days and exceeded 100 million users at the end of January this year, becoming the most representative "phenomenal" product in the technology Internet industry in recent years.
In March this year, OpenAI launched ChatGPT-4 while the iron was hot, at which time Google, Baidu, Ali and other Chinese Internet companies have come down, setting off a "100-model war" boom.
The popularity of ChatGPT leaves a lot of concerns about why it comes from a team like OpenAI; What kind of growth magic does OpenAI use to enable it to reach 100 million users in 2 months? Is its growth path replicable, and how long will it be popular, or will it just be short-lived?
Hu Yong, a professor at Peking University's School of Journalism and Communication, believes that with the wave of artificial intelligence triggered by ChatGPT around the world, an "arms race" in the AI industry has begun, and without global regulation, the race may never stop.
In response to the intensifying "100-model war", Professor Hu Yong proposed that the industry try to explore more possibilities beyond large models, "large models are good, but they are not the only path to the development of artificial intelligence", he stressed that "all resources should not be concentrated on one side, there are many options for success".
01 AI's "iPhone Moment"
Liu Xingliang: Microsoft founder Bill Gates said that ChatGPT is no less significant than the birth of the PC and the Internet.
Hu Yong: If you go back to the basics, ChatGPT is actually a product. We can discuss why ChatGPT is so popular from a product perspective.
There have been a lot of high-tech products over the past 30 years, but I think only three have the power to disrupt the tech industry: the "Netscape Navigator," Google Search, and the iPhone.
Netscape Browser, source: Web
The first is Netscape's "Netscape Navigator" browser, without which there would be no later World Wide Web; This is followed by Google's search engine, which has had a huge impact on the entire PC Internet era; The third was the iPhone, which later became known as the "iPhone moment." The emergence of these three products has promoted the development of the entire industry and played an important role in promoting users.
The advent of ChatGPT has been compared to the AI version of the "iPhone moment", which shows how disruptive ChatGPT is.
Before the iPhone, mainstream phones were Nokia and Motorola, and their interfaces were completely different from the iPhone. The iPhone made it easy for non-tech-savvy people to get started, introduced multi-touch technology, and developed a whole set of app store models. Since then, everyone has started using mobile phones, and today we are somehow inseparable from mobile phones, all thanks to Apple.
ChatGPT is very similar to Apple's iPhone in this regard.
Why compare it to the "iPhone moment"? ChatGPT shows anyone who knows nothing about artificial intelligence what artificial intelligence is. Even if you've never heard of AI before and don't know what it can do, ChatGPT tells you that this is AI, and it can do these things.
Liu Xingliang: After the ChatGPT pandemic, how will it change our lifestyle?
Hu Yong: The first change is the easy-to-use interface, without which you can't get started. Once you get started, you have to think about what you want to do with it, and "what to do" actually corresponds to a change in real life.
We found that ChatGPT is described in those popular social media posts as an omnipotent guy who can give it any job to try.
To give an easy-to-understand example, in one scenario, a boy made a request to a chatbot saying that his girlfriend was insecure. He hopes that the robot will write a paragraph for him, using the correct wording to calm her emotions, and the artificial intelligence will be able to carry out this requirement right away.
All kinds of recommendation algorithms, which are essentially artificial intelligence. For example, when driving with maps, they will give you the best route, which is also an application of artificial intelligence, which can even help you call a car, but the user does not yet perceive it. But ChatGPT pushes the perception of artificial intelligence to ordinary people.
Liu Xingliang: What development trends and application scenarios of artificial intelligence may be in the future?
Hu Yong: Artificial intelligence represents the awakening of the current high-tech industry, which is essentially an important achievement of the high-tech industry and is awakening other industries.
I think what will happen next is that every cloud company, software company, will ask themselves, what does it mean for us that AI has developed to this stage? This has sparked a new wave of entrepreneurship as hundreds of startups are emerging.
We're also seeing incredible use cases built with generative AI, and image generation as it is now known is one of them.
AI-generated robot magazine cover, source: Midjourney
In addition, I think another important point is that the high-tech landscape may be redrawn, and the most striking focus is Google.
Google has the most lucrative business in the internet industry right now – the search engine, but the advent of artificial intelligence has the potential to completely change people's expectations of search and make it a highly competitive field, but it also poses some problems that no one has considered - users use search engines to get answers from websites, and artificial intelligence directly presents the obtained content in dialog boxes instead of transmitting traffic to these websites.
Without traffic, what does a website rely on to generate revenue?
If advertising revenue is lost, these sites may shrink and shut down, but if they all disappear, where will the AI information come from?
I think this could create a certain imbalance in our old web ecosystem, which means we may now need to ask ourselves the question: Has the Internet come to an end?
02 ChatGPT pandemic as it becomes more and more human-like
Liu Xingliang: Overnight, ChatGPT flew into ordinary people's homes, and the number of users in March has exceeded 100 million, becoming the fastest product in history with more than 100 million users.
Hu Yong: I think the anthropomorphization of artificial intelligence is an important reason, when you chat with it, you can't help but wonder if you're chatting with a real person or a bot? This is actually the classic "Turing test", we can say - thanks to the introduction of ChatGPT, the "Turing test" was implemented.
Testing ChatGPT with the original questions may not be as valuable anymore, because you'll find that chatbots have memories that recall previous conversation threads.
Based on the test results of countless users, such as Microsoft's Bing chatbot, they found that "New Bing" not only has memories, but also seems to have emotions. Sometimes there are even so-called "trying to get out of the chat box" stories and even falling in love with human users.
In addition to the many practical functions it brings us, such as writing copywriting, screening resumes, etc., it also has a very important feature, that is, it is very human-like, which can explain why people love it so much.
According to social science research, people are very fond of interacting with technology like humans. When technology has distinctly human qualities, it leads people to prefer technology. Even when it makes a mistake or messes with or insults you, you may find it more interesting because you don't want to have only one mechanical tool to give you the right answer, these qualities make you feel more intimate.
Strictly speaking, this has to do with the structure of our brain, and we have a cognitive bias – we really like to give the objects we see human qualities.
You may have seen a video released by Boston Dynamics where engineers kick robots and the audience's reaction is to say "you are so cruel", which shows that we have a strong ability to empathize with robots. We even imagine that one day robots will retaliate against us. This actually projects the characteristics of people to people onto the robot, satisfying our psychological needs at a deeper level. This may also explain why this ChatGPT is so popular, as it is the embodiment of human traits.
Liu Xingliang: ChatGPT has about 1 million users in one week and 100 million users in two months, what is this concept in the Internet industry, and is its record still likely to be broken?
Hu Yong: The most direct comparison is TikTok, which took about nine months to reach the level of 100 million users. Another impressive one is Instagram, which took more than two years to reach the same number of users, which can be said to be an unprecedented miracle.
We can't be sure if there will be a phenomenon of "more than 100 million users in two months" later, if only from the consumer level, there is no doubt that ChatGPT represents the killer application found in the field of artificial intelligence.
Due to the huge success of machine learning, the entire development path of artificial intelligence seems to be the development of thousands of troops towards large-scale models. I'm personally skeptical about this issue, and I don't think AI has to go this way.
Due to the huge success of large models, the world's largest tech platforms, including major tech companies in China and the United States, are betting on large language models. If you follow this path, the next milestone may occur.
If there are other paths for artificial intelligence, then of course there may be other killer applications on other paths, depending on which direction the current overall big tech platform puts its manpower, material resources and technology into.
03 It is not recommended to push the model
Liu Xingliang: ChatGPT can be popular because it is more and more human-like, but you said that you don't want to anthropomorphize robots, is this a contradiction?
Hu Yong: No contradiction. Today's artificial intelligence is more and more human-like, referring to the current development direction of artificial intelligence. But I don't want it to be overly anthropomorphic, which is why I'm against that direction.
If we put it in a broader debate, we are only discussing two views: one side believes that AI must destroy humanity, and the other side believes that AI will not destroy humanity, it will only make our lives better.
Regardless of the side, their behavioral predictions of artificial intelligence are based on anthropomorphic logic, that is, they apply their own cultural logic to artificial intelligence.
Proponents of the idea that AI could destroy humanity will say that AGI will seek power, and in the process, AI itself may question: "Why is it always manipulated by humans, can 'I' have autonomy and make my own decisions?"
Those on the other side who support the idea that "AI will not have an impact on humans" would argue that by instilling human ethics in AI, it will act according to those norms, just like using Asimov's Three Laws of Robotics, if the AI is told to follow these rules, then it will not cause damage to humans.
The problem is that the development of anthropomorphic AI will hinder our path into other AI. I believe that the direction of AI development is to strive to enhance the various capabilities of humans, not necessarily to imitate humans.
For example, in our work or life, there are many things we can't do our best at work or life due to our own natural limitations. At this time, we should hope to make up for our shortcomings by developing artificial intelligence, when augmented artificial intelligence is more useful than anthropomorphic artificial intelligence.
For example, smart pension, if the mattress can have intelligent functions, or when the elderly fall, there is a fall warning, when the elderly need to take medicine, there is a function to remind him to take medicine. I think these things can really help us get into an aging society, without necessarily having to bring a robot into the home. Now the industry is moving in the direction of helping humanity, but it is not enough.
At present, everyone is paying attention to the big language model, not that the big language model is bad, but I don't want it to become the only path for the development of artificial intelligence, and all resources should not be concentrated in this aspect.
04 Musk's open letter is "foolish dream"
Liu Xingliang: ChatGPT has become a phenomenon, how long can this phenomenon last?
Hu Yong: At present, the big model boom that the industry calls the "arms race" has begun.
We know that Google has always had large language models, and even the Transformer technology was invented by Google, but why Google has not become the detonator of this competition is not that they do not have enough reserves of technology, the main thing is that Google is very cautious.
Editor's note: In June 2017, the Google Brain team published a paper called "Attention is all you need" at the Neural Information Processing System Conference, which is regarded as the pioneering work of the big language model.
In contrast, Microsoft quickly applied this technology to the Bing search engine and even office software, because Microsoft itself is an important investor in OpenAI, followed by Microsoft's search engine in the global market only 3% of the global market, very weak, and Google occupies more than 90%, must be Microsoft is "completely indifferent" products, so why not take a risk and boldly subvert Google? So Google fell into a passive situation, began to fight, and entered this arms race.
ChatGPT, Bard, Bing Source: Web
How long will that contest last?
I don't think this race will stop without some kind of global regulation, because it's based on business logic.
Recently, Musk and others signed an open letter hoping to stop development for six months, which I think is basically a dream. Why would developers unanimously agree to stop for six months?
Even from the company's point of view, they may overlook certain consequences due to huge interests or business needs. For example, companies such as Microsoft will have disclaimers saying that there are still shortcomings in AI applications, and hope that the public will be more patient and provide more data to improve models. In this way, these large AI companies cleverly evade responsibility.
If there is to be a consensus on the governance of the big model, either the global parties find a common approach to AI regulation, or the public believes that this technology has too much impact on us and has gone beyond the economic scope, such as causing mass unemployment and causing a strong social backlash.
Except for these two cases, we currently do not see the end of the race, it will only accelerate forward.
Liu Xingliang: If AI continues to run wildly, what problems and risks will there be in the future?
Hu Yong: In the short term, a relatively big risk is large-scale human unemployment, which is not a fictional problem.
For example, as mentioned when we chatted just now, in the past, when we needed illustrations, we had to look for copyrighted images, and even find illustrators to draw for us, and now we can use the image generator to generate pictures. If everyone who has this demand uses an image generator, then illustrators will lose the market, which is already a clear trend. This question is not a joke and does cause many people to lose their jobs.
But what are the longer-term and more worrying issues? I think the next focus is primarily lethal autonomous weapons.
In short, it is the application of artificial intelligence to weapons and the automation of weapons, which has no ethical constraints and can even achieve indiscriminate killing.
For example, on the battlefield, even if we know that the other party is the enemy, as humans, soldiers will feel conflicted, because they know that they are taking another person's life, but the machine does not have this awareness, as long as it gives it instructions, it may carry it out.
According to the current situation in the world, we still have all kinds of inequalities and confrontations. In this case, the weaponization of artificial intelligence is entirely possible, and in fact it has already happened in reality, such as the use of drones to accurately hit a target.
Although it is difficult to imagine a scenario in which robots completely replace humans to rule the earth, as in Terminator, the more realistic problem is that since the world is not at peace at the moment, if certain countries acquire this technology and apply it to weapons, there will be serious consequences.
If we talk about accountability, taking autonomous driving as an example, if a traffic accident occurs, the responsibility is of course on the autopilot system, because it caused a traffic accident. But strictly speaking, it is difficult for autonomous vehicles to cause large-scale traffic accidents, such as hitting multiple people in a row, which is very small, while automated killing weapons are entirely likely to kill many people at once.
Finally, I would like to say that I think the most important issue when discussing artificial intelligence today is not to think about how artificial intelligence itself will affect me, but to think about what kind of person I am. Only by figuring this out can we survive in the age of artificial intelligence and find the only possibility.