Google PR: An Explainer about Smart Software
January 19, 2023
One of Google’s big wizards packs a brain with the impact of MK 7 16? 50 caliber gun. Boom. Boom. Boom.
Google does “novel” cats. What does Chess.com’s Mittens have to say about these felines? Perhaps, Mittens makes humans move. Google makes “novel” cats sort of move. © Google, 2023.
Jeff Dean has trained his intellectual weapons on a certain viral star in the smart software universe. “Google Research, 2022 & Beyond: Language, Vision and Generative Models.” The main point of the essay / blog post / PR salvo is that Google has made transformational advances. Great things are coming from the Google.
The explanation of the hows of the great things consume about 7,000 words. For Google, that’s the equivalent of a digital War and Peace with a preface written by Henry James.
Here’s a passage which I circled in three different Googley colors:
We are working towards being able to create a single model that can understand many different modalities fluidly — understanding what each modality represents in context — and then actually generate different modes in that context. We’re excited by progress towards this goal! For example, we introduced a unified language model that can perform vision, language, question answering and object detection tasks in over 100 languages with state-of-the-art results across various benchmarks. In future applications, people can engage more senses to get computers to do what they want — e.g., “Describe this image in Swahili.” We’ve shown that on-device multi-modal models can make interacting with Google Assistant more natural. And we’ve demonstrated models that can, in various combinations, generate images, video, and audio controlled by natural language, images, and audio. More exciting things to come in this space!
Notice the phrase “progress towards this goal.” Notice the example “Describe this image in Swahili.” Notice the exclamation mark. Google is excited.
The write up includes Google’s jargonized charts and graphs; for example, “Preferred Metric Delta” and “SuperGLUE Score.” There is a graphic explaining multi-axis attention mechanism. And more.
Enough “catty” meta-commentary.
Here are several observations:
- Artificial intelligence is a fruit basket of methods, math, and malarkey. The fact that Google wants to pursue AI responsibly sounds good. What’s “responsible” mean? What’s artificial intelligence? These are difficult questions, and ones that are not addressed in the quasi-academic blog essay. Google has to sell advertising to keep the lights on and the plumbing in tip top shape… mostly. Seven thousand words is public relations, content marketing, and a response to the wild and crazy hyperbole about OpenAI changing the world. Okay, maybe after the lawyers, the regulators, the content copyright holders have figured out what is going on inside the allegedly open black boxes.
- If the reports from Davos are semi-accurate, Microsoft’s tie up with OpenAI and the idea of putting ChatGPT in Word makes me wonder if Microsoft Bob and Microsoft Clippy will return, allegedly smarter than before. Microsoft is riding a marketing wave and hoping to make money.
- Google is burdened with the albatross of Dr. Timnit Gebru and others who were transformed into former Googlers. What about Dr. Gebru’s legitimate concerns about baked in bias. When one sucks in content, the system does not know that content objects are more or less “better,” “right,” or distorted due to a spidering time out due to latency. The fact remains that Google terminated people who attempted to point out some foundational flaws in what the Google was doing.
Net net: The write up does not talk about Forward Forward methods. The write up does not talk about the likelihood that regulators in the European Union will be interested in what and how Google moves forward. Google is in the regulatory spot light. Will those regulators believe that Google can change its spots like the “novel” cats in the illustration? ChatGPT is something to get venture funders, entrepreneurs, and Davos executives to think positive thoughts. That does not mean the system will deliver. What about Mr. Brin’s self driving car prediction or the clever idea of solving death? Google may have to emulate in part Tesla, a company which allegedly faked the hands-off, full self-driving demo of its smart software. Seven thousand words means one thing to me:
‘The Google doth protest too much, methinks.’ Hamlet, Act 3, Scene 2. (I think Shakespeare put Google in a foul paper and some busybody inserted the name Gertrude.)
Boom, boom, boom.
Stephen E Arnold, January 19, 2023
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
