• Smoking Robot
  • Posts
  • 💥 AI WARS: Google Unveils Plans For Counter Offensive

💥 AI WARS: Google Unveils Plans For Counter Offensive

Google hits back... sort of, Tool of the Day, tech layoffs, mental health and AI, and Links

Author

Kyle Scott
January 20, 2023

Welcome to our new subscribers since the last send. Including you BIG DAWGS. We use only the finest Photoshops around here.

In the newsletter today:

  • AI WARS: Google outlines its counter offensive 💥

  • Tool of the Day 🔨

  • AI has nothing to do with tech layoffs 🌎

  • Mental health and AI 🛋

  • Links 👀

Let the BIG DAWGS eat.

AI WARS: Google Outlines Its Counter Offensive 💥

SmokeBot senses the drums of war in the distance.

Microsoft has dominated BIG TECH headlines over the last few months, thanks largely to a drumbeat of headlines involving their partner OpenAI and its world-shaping ChatGPT.

So awe-striking is Microsoft's hand right now, it has made rival companies' advancements, like Apple's recently announced and insanely powerful M2 MacBook Pros, look pedestrian in comparison.

But now Google has entered the chat.

And by "entered the chat", we mean that CEO Sundar Pichai - Pich-AI? - released a distressingly long 15,000-word(!) treatise on its own endeavors in AI, signaling a counter attack... maybe... at some point in the future... when and if it is responsible to do so... 💤

We read this thing so you don't have to... with the help of ChatGPT.

Here's what Google is working on, in plain English:

  • Google is, of course, trying to solve complex problems to create AI that helps understand, categorize, and generate text, images, audio, and video

  • They are making progress but don't appear to be ready to release any one product similar to ChatGPT, however specific use cases will be spoon fed into existing applications, like Gmail, YouTube, and the Pixel phone 

  • They are working on Chain of Thought prompting, which breaks down requests into individual steps, and on models fine-tuned for specific use cases, like translation and medical contexts

  • They want to create a single model that can work in many different use cases, regardless of context-- this is the hardest part

  • But the most overwhelming takeaway is their insistence on "Responsible AI"--  releasing AI in a way that is not harmful to the world so we're not flooded with biased machines and fake images and video

There's a saying that if you can't describe what your business or product does in a few sentences, it's probably a failure. There's a whiff of that here. 15,000 complex words to basically say "hey, we're not there yet"... all while ChatGPT is accessible and useful to humans with even a moderate IQ.

Google will release a series of blog posts updating their progress... signaling that no major products or features on the level with ChatGPT are expected to be available for widespread use this year.

Not a great look.

Worse is that the New York Times reports today that Google founders Sergey Brin and Larry Page were brought in for Google's "code red" moment last month.

They met with the executive team to discuss AI and adding chatbot features to search, a topic they rarely weigh in on these days.

A slide presentation obtained by The Times outlines a less-than-stellar AI product road map-- as summarized BY CHATGPT!

  • They plan to offer software developers and other companies a list of A.I. programs, including image-creation technology to bolster revenue for their Cloud division

  • MakerSuite will be offered as a tool to help other businesses create their own A.I. prototypes in internet browsers with two Pro versions

  • In May, they plan to announce Colab + Android Studio, a tool to make it easier to build apps for Android smartphones that will generate, complete, and fix code

  • Another code generation and completion tool, called PaLM-Coder 2, is also in the works

  • Google will demonstrate a version of its search engine with its ChatGPT competitor LaMBDA this year, but it's unclear whether it will actually be released to the public this year

  • Google lags behind OpenAI's self-reported metrics when it comes to identifying content that is hateful, toxic, sexual or violent

  • OpenAI bested Google tools in each category and Google tools fell short of human accuracy in assessing content

Meanwhile, today Google laid off 12,000 employees. 6% of its workforce.

Tool of the Day 🔨

R-Y-T-R.

Get it... ryt.

Rytr has one of the best user interfaces we've seen to extract text from GPT-3. 

It provides a set of 40 templates for everything from press releases and blog posts, to ad copy and job descriptions... all pulled from the same API underlying ChatGPT.

But with its easy-to-use interface, users can get the exact result they're looking for without learning a bunch of specific prompts for ChatGPT.

You can read our full review here.

You can t-r-y r-y-t-r today for free right here.

AI Is Not Behind The Recent Layoffs at Microsoft, Amazon, and now, Google 🌎

We read with sharp intent the recent Time story by Andrew R. Chow titled "AI Isn't to Blame for Layoffs at Microsoft and Other Tech Companies." 

Chow's piece addresses panicked fare like this:

Chow took a scalpel to this Tweet and the red herring arguments that underpin it:

"The layoffs at Microsoft and other tech companies have much more to do with current economic conditions than recent AI breakthroughs, experts argue. To put blame on the technology for this wave of tech layoffs would be largely misplaced."

Andrew Chow

Chow then laid out the real reasons for layoffs at Microsoft, Amazon and elsewhere:

  • Many of the jobs lost now were only gained in the first place due to the onset of the pandemic.

  • COVID-19 caused enormous segments of the white-collar workforce to work from home and become dependent on tech (laptops, modems, routers, etc.)

  • Early on, there was a sense that some of the remote workers would never go back to traditional office spaces. They're heading back now, and the need for home tech is shrinking.

  • Plenty of the computer in circulation from two years ago still work great.

  • Tightening economic policy

Chow closed the loop completely with a quote from Syracuse University professor Aaron Benanav: "Because interest rates are rising, there's a shift from the big investors to say, 'No, now you really have to focus on profitability.' And the big way to do that is through cuts."

We already knew that, even if splashy news screeds and alarmist tweets want us (and you) to think otherwise.

Can AI Be Used in Mental Health Field? 🛋

Yuki Noguchi's story for NPR set forth a sad premise: AI may be "the only realistic and affordable way to address a gaping worldwide need for more mental health care at a time where there are simply not enough professionals to help."

But it's not all bad news.

Per the article, which mostly relies on quotes from AI-interested parties and not so much actual doctors:

  • AI is perfectly capable of reminding people struggling emotionally to access basic coping mechanisms like meditation and journaling. No one needs to pay $200 an hour for that.

  • AI can respond to user prompts with appropriate suggestions like taking a walk, taking deep breaths, listening to calming music, etc.

  • As the technology advances, AI is projected to eventually be able to red flag responses from users that indicate worsening conditions that signify a need for human intervention.

  • Additionally, AI can be used in concert with live human therapists to serve as a figurative second set of eyes and ears, catching indicators the human therapist may overlook due to emphasis on other concerns.

We would never suggest that a psych bot could make traditional mental health care and the professionals who deliver it obsolete or even lessen their importance.

But there is a cogent argument that, in the absence of enough trained humans to provide such care AND with that care being prohibitively expensive for so many, a tech stopgap is better than no help at all.

To be clear, the NPR article largely focused on the bull case for AI here. We'll cover this topic more in depth in future sends.

Links 👀