A few days ago, a friend forwarded me a video of Jobs being interviewed in 1981. This video has not been broadcast before, it was only “excavated” in 2021, and relatively few people have watched it. Jobs said in it that, just like the invention and popularization of electric motors, as the cost drops and the threshold for use is lowered, personal computers (PCs) will start from very expensive application scenarios such as scientific research, industry, and military, and gradually enter small and medium-sized enterprises. Personal workers and other commercial scenarios, and then enter thousands of households. He compared the PC to a bicycle. Just like the flexibility and freedom of a bicycle, the PC can become an extension of people’s brains, freeing people from daily tedious and boring work to do more and more creative work.
In 1981, computers were too expensive to be affordable by ordinary families, and most people could not foresee how computers would play a role in their lives. The words Jobs said more than 40 years ago not only predicted the popularization of computers, but also can be used to predict the birth and popularization of smartphones. Today, they are used to predict the role of general artificial intelligence in people’s daily life. Function, I don’t think a single word needs to be changed.
Of course, these are macro forecasts. As for how people will use it, it is still difficult to predict. This issue was discussed in the April 2023 issue of this column.
40 years ago, many people could not imagine that computers would really enter the home, and 15 years ago, many people could not imagine that everyone would have a smartphone…History can only be seen clearly in the rearview mirror, but at the same time, human beings are forgetful . Today, many people still cannot imagine that as the cost drops, general artificial intelligence will eventually become an integral part of people’s work and life, and unleash our creativity.
It is also reasonable to have this idea. The large language model represented by GPT has been popular for half a year. If you are not a technology enthusiast, you may think in your heart, why there seems to be no product that I can use every day? Indeed not. I am a very heavy user, and the browser always puts ChatGPT, Bard, Claude, Pi and other applications that directly communicate with different large language models, but I know that the threshold for using these products is very high today, and I also need to learn some usage skills. I went to an AI conference in an industry two days ago. The most crowded branch was a workshop teaching how to write prompts (which can be simply understood as the skill of talking to AI). It can be seen that even in the technology industry, many people still cannot handle Make good use of current artificial intelligence products. That’s why I don’t think the iPhone moment for AGI has yet to come—I wrote about it in a column in April, and I still think so.
Let alone whether general artificial intelligence can unleash the creativity of ordinary people, it may be necessary to unleash the creativity of practitioners in the technology industry before doing so.
In the past six months, I have been accumulating various product ideas related to large language models while resting. Many of the ideas here are too simple in my opinion, others should be easy to think of, and it should not be complicated to do, but I have been laying flat for a few months and have not seen anyone make it, so I often wonder “Is it right?” Someone already did it, I just didn’t know it.”
It’s not that there’s nothing there, it’s just that it’s not done very well. The new products I have seen in the past one or two months, including the ideas I saw at various demo day events, I feel that the degree of overlap is a bit high, and I rarely see things that stand out. In fact, the so-called bright eyes, my standards are not high. I’m always excited about a product, whether it solves an old problem that others have solved in a creative way, or discovers and solves a new problem that no one has solved before. As for some obvious defects that can definitely be solved, I can automatically complete them instead (sounds like I am a good AI for evaluating products).
For example, there is a product that amazes me called Rewind. After it is turned on, it will record and save your computer screen continuously, as if a person with a super memory is sitting next to you staring at your computer screen. Sounds scary, but you can ask it questions such as “What did I do two weeks ago?” (I just asked when choosing the topic for this article), “My last note about so-and-so Where is it?” (asked just now when writing this paragraph). It is conceivable that ordinary people are worried about privacy, so it should be difficult to promote this product to the public in its current form. But instead I think it’s a solvable problem.
I recently helped an AI startup company do a design workshop, trying to make their products play a more important role in the daily work of the target users, “used every day”. In this process, I think the “classical” product design “artistic ability” is still very useful.
I’m pretty confident about it.
There has been a narrative for the past few years that the “classical” approach to product design is doomed. The rapid development of technology in the AI era is even more so. But I think it is precisely because of this that the “classical” approach is needed.
If there are colleagues who are reading this article, I would like to add that since the “designer” is almost always equated with the look and feel of the product in China, many product design tasks that should be in charge of the designer are carried out by the “product manager”. Done, it doesn’t make sense. The “classical” “product design” job I discuss below is the process of finding a problem that users encounter and coming up with a solution. I call the person responsible for this a “product designer”. I want to emphasize that this is a role, not a position. If you are an engineer but you are responsible for defining the product, you are a product designer at this moment and should have the mindset and skills of a product designer.
And what should a “product manager” do? I mentioned IDEO’s design innovation methodology in the previous column. This is generally the material we introduce design thinking to other companies. Interested readers can download it from their official website for free. Among them, I think the most important point is that innovation must Satisfying human desires, technical feasibility and commercial sustainability at the same time is indispensable. I’m sure the reader can think of some examples of failures due to the absence of one of these.
IDEO emphasizes that innovation should start from people’s desire, which is the so-called “human-oriented design”. But in fact, the excitement that AI brings to product designers, myself included, is that the “technical feasibility” has changed dramatically – a large number of solutions that used to be “not feasible” have become feasible overnight.
But this creates two challenges for product designers.
First, how to understand the technical feasibility of AI? Capability boundaries include both understanding what AI can do unexpected things and understanding the limitations of AI capabilities. At the beginning of the mobile Internet era, designers also faced the same challenges, but the boundaries of smartphone capabilities are still easy to figure out. The capability boundary of AI is a hot research topic even in the AI academic circle. If the product designer is a scientist or engineer, the threshold for understanding may be lower. Even if it is not, to do a good job in the design of AI products, it is as if a painter must be familiar with the characteristics of the canvas, and still needs to know more.
If the product designer does not have a technical background, it needs to be supplemented. I recently started to make products, and I still write codes and adjust them by myself, because only in this way can I understand the boundaries of its capabilities.
Second, after understanding the technical feasibility, it is necessary to re-understand the user’s desire. Creativity is inspired by people’s stories. If it is just to transform existing products, because existing products are limited by the technology before the emergence of AI, there are many problems encountered by ordinary people in daily life that cannot be solved. Just like the transition from a horse-drawn carriage to a car, if you don’t start from people again, it will be difficult to break out of the original product form and create new ones.
Most practitioners may be used to conceiving products from top to bottom starting from abstract macro concepts, rather than starting from specific people and specific stories. But only specific stories lead to inspiration. The above-mentioned workshops for AI start-up companies mainly invited some individuals belonging to their target groups to tell a lot of stories about their daily work scenarios. These people do not need to have used this company’s product. Much inspiration can be gained just by listening and observing how they are doing their work now.
I didn’t use the word “need”. This word is easy to misunderstand, requirement, demand and desire can all be translated into requirements. What is said here is desire. But users like to come up and tell you “requirement” directly, that is, what kind of function they want, instead of telling their own stories and their “desires”. Don’t listen to them. Desire often needs to be obtained through observation, rather than letting ordinary people tell you the products you need. We talked about cars and carriages earlier. Another example, a calculator, a slide rule, and an abacus may appear to be different products, but the problem to be solved is the same.
After the design workshop, the team re-understood the user’s workflow and “desire”, built a prototype of a new version of the product in one night, and gave it to users for trial the next day, obtaining more useful feedback.
You may ask, what is the difference between AI and those outlets in the past? I think the biggest difference is that AI can already play a big role in everyone’s daily work and life today, rather than an imagination. It’s just that the threshold of use is still a bit high. “Classical” product designers are good at solving this problem, connecting people and machines.
This is our job as product creators. But as an individual, if you are interested in the future, I still strongly recommend that you use it seriously.
Being an early adopter of technology comes with a lot of costs, but it can also allow you to see a murky future more clearly. For me, the role of AI products in my daily work can be compared to a calculator – when there was no calculator, the abacus was only used for very complicated additions. One of my daily entertainments in elementary school was to use the abacus to help My father calculates the average test score of each class, and the general calculation can be done by mental calculation; with the calculator, many mental or written calculations are handed over to the calculator; now the calculator can be taken out in the mobile phone at any time, just me In other words, it may have degenerated to only the nine-nine multiplication table and not pressing the calculator.
Similarly, with AI, I can also entrust most of the “brain-consuming” mental work to it. For example, as mentioned earlier, as an amateur programmer who is a three-legged cat, I can write code by myself relatively quickly. I still have to rely on AI to write the code. This is still in the work scene. Imagine that the threshold for using AI is further lowered. Everyone has an assistant with a bachelor’s degree, good general education, but no professional knowledge for the time being. What can it do for you? ? Today I have given AI a lot of mental work that I don’t want to do, and you know that it is not a real person, but you can give it more trivial and more personal work.
As mentioned earlier, there seems to be no products that can be used every day – in fact, you can also try to be your own product designer. The plasticity of the large language model allows you to simulate a variety of functions just by writing the prompt. For example, a robot that helps you write a diary, a robot that teaches you specific things, a robot that prepares ingredients in English, a robot that helps you write Xiaohongshu copywriting, a robot that helps 4-year-old children make up stories…etc.
40 years ago Jobs believed that PCs could unleash human creativity, and 40 years later we believe that AI can continue to unleash human creativity. Human creativity is boundless, and I’m not worried that humans will have nothing to do in the future—many jobs that exist today didn’t exist 40 years ago. This is especially true for children growing up in this era. Take language learning as an example. In fact, in the LLM era, language has become more important. Imagine a child growing up in the age of AI. In the future, he will not have to imitate the format of the model essay writing, but focus on how to express himself in language… Wouldn’t that be great?