Google and Its Big AI PR Campaign
December 9, 2021
I spotted “DeepMind Says Its New Language Model Can Beat Others 25 Times Its Size.” In my opinion, this is part of the Google play to sail forward with its alleged better, faster, cheaper method of training machine learning models. Most people won’t care or know what’s underway. That’s okay because “information” is now channeled through specific conduits. As long as an answer is good enough or the payoff is big enough for the gatekeepers, the engineering is doing its job.
The write up is happily unaware of this push to use 60 percent or “good enough” accuracy to create the foundation for downstream training set generation. But, oh, boy, is that relaxed supervision great for matching ads. Good enough burns down inventory and it allows machine learning models to be trained on content domains quickly and without the friction imposed by mother hen subject matter experts, rigorous analysis and tuning, and retraining using human intermediated data sets.
Plus, skew, drift, and biases are smoothed out or made to go away. Well, that’s the theory.
The jazzy name Retro is not old school. It is new school. The lessons users will learn will take a long time to understand and appreciate its nuances.
This is a big business play and its accompanying PR campaign is working. Just ask Dr. Timnit Gebru, that is the former Google employee who raised the specter of bias, wonky outputs, and the potential for nudging users down the Googley path.
For another example of Google’s AI PR push, navigate to “DeepMind’s New 280 Billion-Parameter Language Model Kicks GPT-3’s Butt in Accuracy.” Wow, just like quantum supremacy.
Stephen E Arnold, December 9, 2021
Comments
One Response to “Google and Its Big AI PR Campaign”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
Your comments don’t make any sense.
“As long as an answer is good enough or the payoff is big enough for the gatekeepers, the engineering is doing its job.” It beats a majority of the industry’s best models on standard evaluation suite.
“But, oh, boy, is that relaxed supervision great for matching ads”. Google is not going to run a large language model in order to do ad matching, this makes zero sense, it’s a straight up categorical error.
“This is a big business play and its accompanying PR campaign is working.” No it isn’t. DeepMind is run fairly separate from Google, and their work on AlphaGo, AlphaFold, and other research is years old, and has no business revenue currently. Demis Hassabis, when he got acquired by Google, negotiated a fair degree of independence.
“Wow, just like quantum supremacy.”
And what’s wrong with touting your research achievements, like every other company? It helps with recruiting, because scientists and engineers want to work where the most progress is happening. And Scott Aaronson, perhaps the leading expert on Quantum Computing these days, pretty much confirmed Google’s claims on Quantum Supremacy.
Now, a true quantum computer won’t exist until one can get a few hundred *error corrected* qubits (100,000 – 1 million regular qubits), but you have to start somewhere, and a proof of concept of quantum supremacy, no matter how useless, helps reduce risk and increase confidence for other investments to follow, the same goes for EVs (eg the Roadster), or the first high temperature super conductors, or the first EUV lithography prototypes. What NIST demonstrated 20+ years ago in the lab, no matter how primitive, proved that ASML’s investments would eventually bear fruit.
Pretty much no one in the top AI field is pooh-poohing DeepMind’s achievements here. But for some reason, you’re salty about it to the point of incoherence.