For a few weeks now, we’ve seen Google shake before the power of AI.
Despite being trained only with information that goes as far as 2021, ChatGPT has sometimes proven to be more friendly and efficient than the now seemingly old-fashioned Google search.
About 58% of Google’s revenue comes from Google Search, so this is making them very, very afraid and pushing them to consider desperate solutions.
First of all, they want to implement generative AI directly into search engine result pages. This is huge!
It means that the average Joe will have a well-digested concise answer for any question he has. However, this also gives Joe fewer reasons to visit your website, blog, or any other digital activity you have, rendering all that SEO useless.
One can only imagine how advertising will be integrated into this model.
But implementing generative AI would mean nothing if it can’t compete with GPT-4.
That’s why Google unveiled PaLM-2, a language model available in various sizes: Gecko, Otter, Bison, and Unicorn. Each size offers different trade-offs between speed and sophistication.
Google seems to view Microsoft’s Co-Pilot as a menace, prompting them to bring AI to all their business applications like Google Docs or Google Sheets.
It’s called Duet AI, and while it may not be as good as GPT-4 currently, it has the potential to improve, especially with the Bison model of PaLM-2. It could outperform GPT-4 once the Unicorn model is implemented.
Oh, and did I mention that these models can also write and debug code?
Lastly, there’s Project Tailwind, a language model that can be trained with all the data you upload to the cloud for storage. This potentially opens the gate for a 100% customizable personal AI for every one of us.
I don’t know about you, but it seems to me that Google is putting plenty of eggs in the AI basket. Their efforts reflect a determination to embrace AI across various domains and adapt to the evolving landscape of technological advancements.
But will the gamble pay off?Â