- For the first time, in twenty years, Google was threatened at the point, of summoning up Page and Brin back to the company, to figure out an AI strategy which moves along, launching over 20 AI-based products within the Google’s ecosystem. These might go from simple product enhancements to search, voice and productivity. Up to the launch of a ChatGPT-like product, anytime in 2023, called Sparrow, which might be an AI-based search assistant, that might be more grounded, factual and able to cite sources.
- Google AI strategy. In the meantime, the launch of ChatGPT seems to have awaken the AI tech giant. As Google has led the way in AI, being the first company, which in 2018, announced the transition of Google, to become an AI-first company. This week Google published its manifesto for AI. And it also published a research into the AI. Let me though tell you the most important part of it. Google explained that it’s working on multimodality (generative models able to handle anything, from text, images, audio, video and more). Multimodality, if achieved, can be what would enable Google to get back on track, to the race of AI!
- In the meantime, it seems that OpenAI is already moving forward with a premium version of ChatPGT! It seems that OpenAI is moving forward with a very simple pricing structure (for now) where the premium version of the product might be priced at $42 per month. To understand how this fits into the OpenAI business model, read this!
- This week, Satya Nadella announced how OpenAI’s products are getting integrated within all of Microsoft’s business model! To understand how the whole thing is playing out read the OpenAI/Microsoft partnership structure here.
- AI and new job skills: The most exciting aspect to me, of this AI revolution resides in the fact that this current paradigm, is getting achieved through a new architecture, called transformer-based and scale. Indeed, once this new paradigm has been tested, the remaining part of it (beyond a few new minor techniques) was the result of scale! In the last six years, by using more data, in the pre-training phase, more parameters, and by training these models with better data, or for longer, we achieved incredible outcomes! For how long, and what scale can still do, on the current paradigm is very hard to say, but we’ll see. In the meantime, the AI industry is already creating new types of jobs that didn’t exist before. One example? Prompt engineering; and don’t get fooled, as this isn’t necessarily a technical role (indeed it requires minimum coding experience for now), it’s a type of job, that is more closer to product development, than programming. In fact, my main argument is that, since coding might get – in part, commoditized – via new AI coding assistants, what will matter will be the intersection of technology, product, and distribution, to enable network effects into AI products. How much does a prompt engineer make? According to a job posting this week, it seems anywhere between $250-330K per year!
- Are we in an AI Hype Cycle? Yes, we are! I’m aware that we’re going through a massive Hype Cycle for AI. Indeed, we’ll see what AI will be able to achieve, and what will be its major limitations, as we go along. In the meantime, as business people, it’s critical to draw the lines, between being skeptic (understanding limitations, drawback and dangers of AI) vs. being cynic (look at it as if it’s all hype, without removing the noise and take the signal in).
- New conversational search engines are springing up like mushrooms. A first nice release was that of Perplexity AI, a conversational search engine, ChatGPT-like, which though is able to provide sources and references as it generates content on the fly, based on the journey of each user!
