📕 Google Releases Its AI

Job losses, new AI, and fake images of Trump. Oh my.

In the email today:

• OpenAI releases list of jobs most at risk to AI 📉

• Google releases Bard 📕

• Fake Trump images dominate Twitter ⚖️

• Headlines 📰

• Links 👀

All at once. Here we go.

OpenAI, University of Pennsylvania Author Study Identifying Which Jobs Are Already at AI Risk 📉 

Amid the rampant enthusiasm surrounding AI lurks an underpinning of anxiety caused by fear of widespread job losses.

You might think OpenAI would look to tamp down this anxiety. Apparently not.

Here are the specifics in quick hit form:

• GPTs have the potential to significantly affect a diverse range of occupations within the US economy, with higher-paying jobs being more likely to be automated.

• Jobs heavily reliant on scientific and critical thinking skills are less prone to automation.

• Jobs requiring proficiency in programming and writing are more susceptible to being automated.

• Specific jobs at high risk of automation include Web and Digital Interface Designers, Blockchain Engineers, Tax Preparers, Writers, and Authors.

• The study also identifies 34 jobs with significantly lower exposure to GPT-powered models, including athletes, cooks, mechanics, helpers, tire repairers, etc.

• OpenAI CEO Sam Altman has acknowledged that ChatGPT could eliminate jobs but also create new ones that are hard to imagine today.

The study reinforces what we already know about what AI can do and what it cannot. Occupations requiring physical human skills are mostly beyond AI’s capabilities (for now). No robot is rolling up to your car to change your flat tire roadside, or showing up to your home to unclog your toilet.

It is fascinating, though, to see that there are a lot of white-collar, high-paying jobs directly in AI’s crosshairs. Accountants, lawyers, writers (gulp)…they’re all in trouble because so much of what they do is either repetitive or derivative of work that has all be done before.

So weirdly, this Ivy League-inflected story should probably convince a lot of college-bound teenagers to opt for trade school or at the very least an undergraduate program with a heavy technological bent.

Because it’s pretty clearly well past time where majoring in English literature will qualify you to do anything other than tend bar after you graduate.

A Day Late and A Dollar Short, Google Releases Its Admittedly Imperfect ChatGPT Copycat, Bard 📕 

Google was so dominant for so long in the tech world that watching Google struggle as AI explodes feels a little like watching Michael Jordan limp around with the Washington Wizards.

Bard, Google’s purported “ChatGPT Rival,” is out in beta, but even reading the story about the release gives SmokeBot a bit of a sinking feeling. And SmokeBot ain’t the only one.

These quotes and observations from Google’s braintrust, they’re just not what we’ve come to expect.

• CEO Sundar Pichai, as Bard is being launched, admitted to employees that the program would need to improve through user feedback and that “things will go wrong” with Bard initially.

• The Google blog post trumpeting Bard’s arrival included beauties like “while LLM’s are an exciting technology, they’re not without their faults” and “they can provide inaccurate, misleading or false information while presenting it confidently.”

• Perhaps most damning is Google’s admission that Bard is, at this point, an “early experiment.” Kids, it’s way too late in the game to be in the early experiment stage.

SmokeBot hates to repeat itself, but here we have another instance of a product being released quite obviously before it’s ready to be great in the interest of being sort of good enough right now.

Google missed an opportunity here. If Google had released Bard and said “it’s ChatGPT without all the hallucinations and glitches,” that would have been a story.

Instead, Google released Bard well past the time that Bard’s capabilities were at all novel or compelling while also admitting that Bard is as error-prone as the thing it’s clearly copying.

Oh, Google. Pull yourself together.

AI-Generated Wish Visions of Donald Trump Being Arrested, Though Amusing, Also Augur Trouble ⚖️ 

Reports on the possible/probable/unlikely (depending on the source) arrest of Donald Trump continue to percolate in the news cycle.

Unsurprisingly, there are many people who absolutely cannot wait for this to happen, so much so that they are taking to AI to generate fantasy images of what that scene might look like.

There are many other such images floating around online, including depictions of Trump behind bars in an orange jumpsuit and (perhaps most absurdly) Trump engaging in hand-to-hand combat with a police officer on the streets of New York.

As weak as these images are, though, they still foretell serious trouble as we gingerly enter another election cycle.

Sophisticated eyes like SmokeBot’s can see right though the artifice in these images. But elections aren’t generally won by securing the votes of only second- and third-level thinkers. Plenty of normal, trusting people could look at one of these images and conclude, at the least, “well, maybe?”

We’ve not seen the last of this sort of thing. Not by a long shot.

Headlines 📰 

Bill Gates says AI is the second truly significant technological advancement in his lifetime: SmokeBot bypasses the vast majority of the blowhards who overstate AI’s probable impact on our lives, but when Gates speaks, we listen. 🍎 

As some companies embrace AI, others ban it: Time will tell whether this is a prudent choice or ignorance, but some companies are prohibiting employees from using ChatGPT and its offshoots at work. 🤷‍♂️ 

Coca-Cola leverages AI to create consumer engagement: Through March 31, Coke enthusiasts (not that coke) can use the beverage’s AI platform to “Create Real Magic,” and some of the images those consumers create may end up on digital billboards in New York and London. 🥤 

Nvidia seeks to democratize business AI, at a cost: “Chief Executive Jensen Huang on Tuesday laid out the company’s plans to make…technologies like ChatGPT available for rent to nearly any business.” Any business that can afford $37,000/month, that is. 💰️ 

Links 👀

• We’ll start hopeful, as some of the links below are dire, but if you’re living in the deep Australian Outback, AI could save your life 👍️

• By contrast, the good people (and some of the bad ones) of San Francisco apparently believe that AI is going to kill us all ☠️ 

• The truth is always in the middle, so maybe the best idea isn’t rejecting AI but just pumping the brakes 🛑 

• Anyone who has ever driven over a swing or lift bridge should appreciate AI’s capacity to keep everyone out of the drink 🌊 

• No surprise that the biggest insurance companies are deploying AI to try to stop workplace accidents before they happen 🏥 

• Screenwriters are already willing to compromise with AI to save their jobs ⌨️ 

• Ever the firebrand, the Republican Party looks to create a right wing chatbot 🙄