Newsletter

Exploring the Essential Value of Generative Artificial Intelligence: A Conversation with Kento Kajitani

You can listen to the “SZ MEMBERSHIP PREVIEWS” podcast here.

────Hello everyone. At PREVIEWS at SZ MEMBERSHIP, the editor-in-chief will decipher the most popular stories among the articles published for SZ membership. This week’s theme is “AI” and today we have a special guest, Kento Kajitani, representative of POSTS, who will release his book “How to Create a Company/Organization that Will Win in the Age of Generative Artificial Intelligence” on February 15th . .

Michiaki Matsushima (hereinafter referred to as Matsushima)Kajitani took the stage at the session “The AI ​​World In 2024: Re-introduction to Generative AI” at the “WIRED Futures” conference held late last year. In view of the release of a new book that delves into the contents, I would like to get closer to the essential value of generative artificial intelligence by linking it to selected articles. First, in this week’s selected articles, we publish two articles about the film “Her”. It’s been more than 10 years since it was released, but have you seen it, Mr. Kajitani?

Related article:AI will no longer be a “philosophical zombie”: AI and humans from 2024 to 2050 as discussed in WIRED Futures[Video archiviato incluso]

Kento Kajitani (hereinafter referred to as Kajitani)I saw it a while ago. It depicts a world where the main character forms a deep relationship with an interactive artificial intelligence, but looking back it seems more like a film depicting “now” than a science fiction film. It’s work that makes you understand how much the way we perceive AI has changed.

MatsushimaIt’s exactly true. One of the selected articles is an essay that was published in the technology media “Ars Technica” and, as it is written in the article, it is said that everyday life depicted in the film has arrived, such as AirPods and conversations with ChatGPT. I feel. Another WIRED article stated that while the cityscape with its skyscrapers and the main character’s standard of living are still fantasy, the future in artificial intelligence has truly arrived.

This week’s articles:Are we living in the era of the movie “Her”?

KajitaniCertainly. It’s a little ironic that it’s just software related.

MatsushimaYES. The main character is a letter writer. In a world where AI is so advanced, I think it would be fine if AI wrote for us, but humans can also make a living doing it as a profession. This article is interesting for its critical perspective.

KajitaniRight. It’s hard to talk about “Her” without revealing spoilers, but I think it’s an interesting movie with scenes where the human ego and the true nature of artificial intelligence contradict each other. The main character tries to make the AI ​​his own, right? But as I said at the conference, one of the essences of generative AI is that the marginal cost of generating content can be reduced to zero. There are humans who want to make artificial intelligence their own and an artificial intelligence that can interact with many humans at the same time and generate unlimited content. I think the contrast is drawn beautifully.

MatsushimaKajitani gave an in-depth talk at the conference on the essential value of generative AI, and the culmination of his discussions is his new book, “How to Build Businesses and Organizations That Can Survive in the Age of Generative AI.” ‘ Congratulations, you’ve decided to reprint the book before it goes on sale!

KajitaniThank you. I believe a growing number of people are experiencing a sense of crisis and opportunity around the impact of generative AI, but it’s unclear what specifically they should do when trying to reform their organization in a positive direction. on this topic, so I decided to write one myself, putting all my experience in business management, new business development and business consultancy work in the field of generative artificial intelligence.

MatsushimaI believe you have introduced seven essential values ​​of generative AI in your book, but could you introduce some?

KajitaniYES. The first is to “reduce content creation costs to zero.” I mentioned it before. The second is the “realization of an infinitely natural dialogue through the system”. I think a lot of people have also experienced this using ChatGPT.

MatsushimaIn this regard the film “Her” also got to the point.

KajitaniI agree. The third is “flexible processing by vectorization of unstructured data.” This is a point that is often overlooked. In other words, until now, when creating an AI-like system, it was first necessary to organize the data. However, when using a large-scale language model like GPT, which is the core of generative AI, it converts input text into arrow data called vectors, so even if the data isn’t organized at all, you can just dive into it . He somehow handles it in a nice way. This is truly innovative and a service called “Glean” has been created, a next generation internal search engine, which is playing a role in sharing internal knowledge.

Other areas include “multimodalization of content,” “democratization of high-cost specialized knowledge,” and “realization of new input methods.” For more details, see the book, but “reduce language barriers”. ”That’s the interesting part. A large-scale language model like GPT translates all languages ​​into an AI language and stores it. At some point, I think there was a time when the artificial international lingua franca, “Esperanto,” became a hot topic. This converts English, Japanese, and Korean inputs to vector data, which is the direction and length of an arrow. There are no linguistic differences within GPT. This makes it easier to translate and generate documents between different languages.

MatsushimaWIRED Futures also aired a video of Mr. Kajitani speaking fluent French. Rather than saying that translation technology has improved, the improvement is because they are processed using the same thing.

KajitaniYES. Therefore, the speed at which global services support Japanese is becoming much faster.

MatsushimaI see. Also, I was impressed by your talk about “AI neocortex” as a new layer of the brain at the conference. The content was that artificial intelligence is not something that exists outside of humans, but something that is inside them.

KajitaniThank you. As artificial intelligence evolves, what will happen to the meaning of human existence? I am often asked this question and my answer is: “I think we are entering an era where humans and artificial intelligence should not be considered separately.” This will lead to the creation of an “AI neocortex” as a new layer of the human brain. As generative AI continues to evolve, for example, there will be differences between engineers who use it and those who don’t. that there is already a 10x difference in coding productivity. It is clear that the difference will increase 100, even 1,000 times, and could probably become the difference in ability between new humans and old ones.

There has been talk recently that Neuralink, led by Elon Musk, performed brain implants on human patients for the first time, and that improvements in brain-machine interface technology will make it possible to directly connect artificial intelligence and the brain. that we will no longer think only with the human brain, but will also include the processing power of artificial intelligence as a layer and consider that processing power as part of the individual’s capabilities.

I don’t think this is a far-fetched story; sometimes we still check your computer skills during interviews, right? We evaluate a person’s capabilities, including whether he can operate a PC that expands his capabilities, and then we hire him. I think it’s a continuous change as the contour gets broader.

MatsushimaIn addition to describing the worldview so far, the book also examines how society as a whole will change over a 5-10 year time frame and also introduces representative future forecasts for each sector. Also, in the last chapter, you talk about “generative AI native organizations” and I think a lot of people will pick up this book because they want to aim for that.

KajitaniWe are thinking about AI-native generative organizations in two phases. First of all, this book clarifies through several research findings that there is a productivity difference between individuals who use generative AI and those who do not, so it will become a group made up of individuals who are capable of using generative AI. as the first phase. The other is structural change. In an era where tasks that were once performed by 100 people can now be performed by a single person, traditional organizational systems and structures will no longer work. In short, the question is how to transform it into optimal form in the era of generative artificial intelligence.

For example, there is no doubt that the trend towards smaller teams will continue. At that time, middle managers who served as information transmitters could be invited to become protagonists by learning new skills, or they could be used as hubs to manage the growing number of small teams within the company. to connect… I believe how to change this structure and employee skills is one of the main topics, so I will write about it in advance.

It has a specific roadmap and framework, so if you follow it, I think you’ll become a pretty strong organization. As a further step, we also summarize the future organization that is optimal for the era of generative AI. So please give it a try Attempt.

────It seems like this is a book where a variety of people, not just executives, can get tips. Thank you so much, Kajitani-san! February WEEK #1 includes “‘OpenAI will never be number one’ – Yann LeCun talks about the future of artificial intelligence that has gone ‘open source,'” “The ‘ability’ of story-generating AI to put to bed the children” and more., the seventh installment of the “For Creators” series has been released, so please check it out.

[Ascolta la versione completa con audio! Attendo con ansia anche l’angolo WIRED RECOMMENDS! ]

*This article is not a transcription of the audio, but has been edited for length for ease of reading.

(Edited by Erina Anscomb)

#essential #generative #artificial #intelligence #represented #science #fiction #films #Newsletter #VOL.222AI #WIRED.jp