Google AI Thrashing: A Fresh Sundar & Prabhakar Comedy Script?
May 5, 2023
I love anonymous information from a Discord group. How about those Discord app interfaces? Aren’t they superior to the DEC PDP 8 command line? Plus, the folks publishing this internal document (allegedly from the Google control room) do not agree with the contents of the allegedly leaked and not stolen document. That’s reassuring to some I assume. So let’s look at this glimpse of the sausage being made.
I read “We Have No Moat and Neither does OpenAI” on the SemiAnalysis.com web page. The write up is, like most “real news” arrives with a soupçon of mystery. The subject of the information is that Google is addled by OpenAI and the zippiness with which some online users are innovating, applying, and de-Googling certain of their activities.
I noted this statement:
Things we [presumably the happy band of unified Google smart software wizards] consider “major open problems” are solved and in people’s hands today.
And this assertion:
Open-source models are faster, more customizable, more private, and pound-for-pound more capable. They are doing things with $100 and 13B params that we struggle with at $10M and 540B. And they are doing so in weeks, not months.
I think this means old and slow, not young and snappy. The force in the green fuse at what may well be the Google is mired in the Mountain View digital La Brea Tar Pit. Not good for Googzillas young and vibrant or old and arthritic I would suggest.
I liked this statement which seems a bit un-Googley to be frank:
The barrier to entry for training and experimentation has dropped from the total output of a major research organization to one person, an evening, and a beefy laptop.
Beefy laptop? What? No Chromebook?
But aside from jargon, what’s Google big advantage? The write up notes:
All this talk of open source can feel unfair given OpenAI’s current closed policy. Why do we have to share, if they won’t? But the fact of the matter is, we are already sharing everything with them in the form of the steady flow of poached senior researchers. Until we stem that tide, secrecy is a moot point. And in the end, OpenAI doesn’t matter. They are making the same mistakes we are in their posture relative to open source, and their ability to maintain an edge is necessarily in question. Open source alternatives can and will eventually eclipse them unless they change their stance. In this respect, at least, we can make the first move.
This is an interesting “upside” to the analysis. Look OpenAI will die, but possibly an outfit like Google can get a win with an open source approach. I am not sure I can agree when many people are happily embracing Microsoft’s crowd-pleasing approach to smart software. Hey, it’s Windows. Edge makes life easy. Outlook and Teams will be even more wonderfulest with Microsoft’s smart software.
The open source angle is interesting, but I am not sure it will resonate in the current hot house of smart software where genetic diversity is increasing with each karyotype mash up. This is not elephant gestation. Think swamp critters on a warm spring day.
Am I citing information from what is allegedly a leaked Google document? I don’t know.
The write up is not Googley as documents from the era of Messrs. Brin and Page (where is that wizard because some in a judicial system have a document for him), and that adds to the comedic vibe. Nope, the essay not ready for the Sundar & Prabhakar Comedy Show on YouTube this weekend. It’s a good start though, just behind schedule.
Stephen E Arnold, May 5, 2023