Product x Generative AI 1/5:'IA générative sera l'interface principale avec le monde numérique
Last week, Homeric achieved a significant milestone by delivering its first public webinar, which exceeded our (reasonable) expectations for a first event of this type. We want to thank all the participants who joined to learn more about how generative AI will change product development and management.
During the presentation, I planned to share important messages that will enable innovation in the era of AI. However, I came to the realization that these messages are deserving of their own dedicated webinar. As a result, I decided to articulate my thoughts on these insights and share them. I believe that if these insights are helpful for me, they will likely be helpful for you too and especially the product leaders, managers, designers, and engineers out there.
Generative AI (Gen-AI) will change the nature of digital products and how we create them. We cannot fully comprehend the ways in which the technology will change digital products because it is still in its early stages. However, some principles are and will remain true in the future, and they are enough to explain that Gen-AI is poised to play a central role in our lives:
Gen-AI will become the main interface with the digital world
Gen-AI's natural behaviour is imperfectly perfect
Gen-AI comes with an infinite number of built-in features
Gen-AI & Regulation - My take on the the Yoshua Bengio vs Yann LeCun debate and why it feels personal to me
Gen-AI & Product Development - How product management, design, and engineering will change forever
Today, let’s dive into our first topic:
Gen-AI will be the main interface with the digital world
Many people claim that AI is the biggest technological advance since mobile, Internet, or the Personal Computer, but I think the best comparison is to the graphical user interface (GUI). AI assistants and co-pilots will become the primary way to interact with the digital world because they’re much more convenient than the window/icon/mouse/pointer approach we’ve been relying on for 80 years. It was great validation to see Bill Gates use the same example:
"I knew I had just seen the most important advance in technology since the graphical user interface (GUI).”
Bill Gates, The Age of AI has begun - https://www.gatesnotes.com/The-Age-of-AI-Has-Begun
Many digital constructs we think as “normal” are actually very basic for our brain
It’s almost a shock to realize how we’ve been conditioned to the concepts of folders, directories, and lists. Windows, Google search, documents, file systems, sitemaps, and most of the digital things we’ve built over the last decades are objects that can be listed, linked, and sorted. That era dies with Gen-AI.
While pretty effective, these experiences are not optimal and not aligned with our natural thinking. With Gen-AI, we no longer need to search for something on Google. Instead, we can just ask Chat-GPT, which immediately finds what we need.
On our human side, when we think of our memories, for example, they are not listed in various folders in our brain. We don’t access them by opening a drawer, listing all our memories about an event, selecting a memory, and then consuming it. We just think of it, and it comes (or not, or partially, but these variations will be discussed in the “imperfectly perfect design” article 😉)
Gen-AI is much closer to our natural way of accessing information. We think and communicate something we want, and the AI will go and get it back to you. It’s simpler and faster, and if the AI is good enough, making the response relevant to you, it just makes everything else obsolete.
For these reasons, AI assistants and copilots are meant to drive higher engagement rates. The insane traction Chat-GPT demonstrated as well as some interesting insights from VCs assessing Gen-AI pre-seed application with user engagement rates so high they would be considerer as outliers with regular applications.
Gen-AI will be the all-in-one unique window you need to interact with the digital world
Digital products deliver value through images, text, and sounds. AI already demonstrates impressive capabilities around images and text thanks to models like Midjourney and Chat-GPT. The audio part is a bit behind, but many AI researchers will tell you this is the next big leap for Gen-AI. Once you have an AI assistant able to generate any image, text, or sound you need when you need it, you can craft any digital experience on the fly. The consequence of this paradigm shift is breathtaking:
You will not browse the Internet anymore. The Internet will morph around you.
The AI will generate the digital asset you need, even if it didn’t exist beforehand.
Sitemaps and websites will be reinvented as well. Why would you browse different web pages when Gen-AI can just generate a personalized web view you need without having to navigate to another page ever? For example, when I visit SaaS websites to find a new tool, I quickly scan the homepage for the hero section and the titles. I go to the pricing page to understand what budget I’d need and read the feature comparison table. At some point, my AI assistant could just generate a view just for me with the Hero section and the pricing table. Perfect homepage for me, and as I scroll down through this hyper-personalized web page, the AI, following my instructions or predicting the next section I need to read, could keep building the page as I want to dive deep into the details of the solution. The Gen-AI assistant becomes permanently positioned between me and the digital world.
A continuous stream of information
As of 2023, Gen-AI and especially LLMs don’t think. Even if they can generate mind-blowing outputs, the only thing they do is predict the next word of a sentence (it’s a bit more complex, I know, but this is the insight I think everybody should be aware of and keep in mind). This is why there is this streaming effect when we use Chat-GPT; we’re witnessing the continuous generation of the next word, and the response is built in front of us in real time.
Code is text-based
LLMs generate text so they can continuously stream HTML code to be displayed
→ Gen-AI can build in front of our eyes a never-ending web view that will present everything we need to know.
I could seamlessly scroll from a section about the weather forecast to a New York Times article, to my email inbox, to the photos of the resort I could book for my upcoming family vacation. The concepts of URLs, apps, websites, finite documents, or assets don’t really make sense in a world of raw data flowing to the AI assistant so it can generate the right view for the user continuously and almost in real-time.
As an interesting corollary, we could wonder if this will not go against some core principles of Web 3.0, blockchain, and especially NFTs:
What’s the value of a finite digital asset like an image, a 3D model, a code function, etc., if Gen-AI can create an infinite stream of better alternatives in seconds?
How do we keep the concept of a blockchain token for digital assets? (it remains super valuable for physical assets).
These are open questions, and your thoughts about them are more than welcome. My insight here is that Gen-AI is significantly disrupting Web 3.0, and that field is poised to go through many other turbulences before going mainstream if it ever does.
We may experience the entire digital world through the ultimate window of a Gen-AI system.
We may not move around the Internet anymore. The Internet will move around us.
Digital products may no longer depend on digital properties and assets produced by the solution provider during product development but rely on custom hyper-personalized Gen-AI renderings generated downstream when delivering the experience to the user.
Gen-AI enables a continuous, never-ending digital experience. Will we experience the Internet as if it was continuously streaming to us?