Good morning, this is Tools of AI- we bring you about all the happenings in the field of Artificial Intelligence, piping hot, just like your morning coffee.
Anyway, Google CEO Sundar Pichai aka – the guy every brown kid is told to be like- just announced that Google will be rolling out their own Chat GPT competitor over the coming weeks and months.
He mentioned this during an earnings call for Q4 2022, just a few days ago. He said that google would be making their search interface much more conversational and would utilise their existing LaMDA model (Language Model for Dialogue Applications).
For those who haven’t heard about it- you might remember a news story from 2021 where a Google engineer claimed that Google had created a sentient AI that had racist, sexist, and a few other -ist based tendencies?
He was talking about LaMDA.
Anyway- Google has been hard at work to iron out the kinks in their model before revealing it to the public. It has been reported for a while now that since Chat GPT-3 launched, this project has been moved to the top-most priority.
Makes you wonder why Google didn’t roll out such a feature much earlier right? Why did it have to be forced by Open AI to innovate?
One potential reason is the inertia of being a humongously large organisation. The inertia isn’t just lethargy based, it’s also ideology based.
Google’s employees have in the past taken very vocal stances against some of it’s projects and the way they were being executed. Case in point- Google had to withdraw a few years ago from a Department of Defence project because it’s employees were against the notion of developing technology that could potentially be weaponised.
The other thing is- Google has a lower embarrassment threshold and has much more to lose if it’s AI models give out the wrong results or tend to be caught being even a tiny bit left or right wing. Open AI has been receiving a lot of criticism in this regard over the past few weeks, however, being a start-up, they can roll with it and improve.
For Google, however, things aren’t the same- which is why they seem to be extra careful before rolling out any new features.
How Good is Google’s LaMDA Model ?
While no one can say for sure until it’s release, a few people on twitter that have gained paid developer access to the model say it’s just…meh, compared to Chat GPT-3.
I’m sure that Google’s executives would know that they need to consistently provide at LEAST comparable performance to Chat GPT, if not better it, in order to retain it’s position as the #1 search engine.
Google has a lot going for it- they have an unthinkable amount of data collected from google search users over the years, as well as the resources of a mammoth organisation behind themselves.
However, as we have seen time and time again, David beats Goliath because it was laser focused on just one thing. Goliath, being too big, moves too slow to notice the important things.
This has happened in tech over and over again, but the speed at which things are changing right now can give anyone whiplash.
(And that’s why you have to read Tools of AI).
Talk next week!