😺 Yikes, Google

PLUS: racially diverse nazis...WTF?!

Welcome, humans.

Political consultant Steven Kramer confessed to funding the AI robocalls that mimicked Joe Biden discouraging people from voting. Kramer said more enforcement is needed to stop “people like him from doing what he did.”

Uhhh, that’s like getting caught stealing a donut from Krispy Kreme and saying, “there needs to be more security to catch thieves like me”. He’ll need to pull a rabbit out of a hat to explain this logic!

Here’s what you need to know about AI today:

Google faces intense backlash over Gemini's odd responses.
Perplexity launched a new podcast that’s AI-generated (& it’s good).
Android users: ChatGPT widgets are now available for your home screen.
We found a quiz to see if you can spot AI-generated videos.

Google shoots itself in the foot.

Last week was the best AND worst week for Google re AI.

The good news is that its upcoming Gemini 1.5 Pro model showcases remarkable capabilities with its expansive context window (details forthcoming).

The bad news is Google’s AI chatbot “Gemini” is getting A LOT of heat after generating some outrageous responses. Take a look:

We haven't seen this many founding fathers depicted as people of color since our third Hamilton marathon on Disney+ (it never gets old)!

But seriously, how the heck did this happen?!

This thread does a pretty good job explaining the mishap:

TLDR: Google gave Gemini special instructions to be “diverse” → Gemini took the instructions a bit too literally → the results were *bonkers* → & it’s a reminder to ensure AI aligns with what we actually want.

So Google apologized, turned off its image model, and said, “we’ll do better.”

For Google, the silver lining is these issues are super fixable (throwback to ChatGPT’s own biases). Let’s just hope it doesn’t take as long as it did to release Gemini in the first place!

FROM OUR PARTNERS

Hyperautomation: Not Just for Tech Wizards Anymore!

Picture this: It's 2024, and everywhere you turn, people are telling you to let AI do your work.

But how? Tech wizardry isn't exactly your forte.

So you use AWS Hyperautomation.

In this bold new world, there’s no such thing as “prompt engineering.” You just tell your AI Hyperautomation tool what you’re looking for in plain language, and the tool gets to work.

It’s like “talk to me, friend” to “I got you; here’s the workflow” in a click.

Learn more about the future of automation with AWS.

Perplexity's AI podcast offers a glimpse into the future...

Ready for something wild?

Perplexity just dropped this new podcast, Discover Daily, that recaps the news in 3-4 minutes.
It already broke into the top #200 news pods within a week.
AND it's all *100% AI-generated*.

The podcast uses ElevenLabs to transform trending news on Perplexity into fully-produced podcasts with synthetic voices. Our take?

It's surprisingly solid for a news podcast. Concise. Unbiased. And the voice is nearly as soothing as David Attenborough's.

The only hiccup? The podcast doesn't disclose it's AI-generated.

Why it matters: While we're fans of podcasts with dynamic hosts, Discover Daily presents a novel approach to producing entire podcasts with AI.

What's next: If Perplexity can pull this off, what's to stop them from making a Discover Daily tailored to each listener's preferences for topics, length, style, and voice?

Personalized podcasts for billions of people? Perhaps we're jumping the gun.

P.S. Use ElevenLabs or Podcastle to convert text into podcasts!

Around the Horn.

HAHA

Lenny Rachitsky showcases 20 ways teams use custom GPTs to boost productivity.
*advertise in The Neuron and put your company *right here* in front of 400,000+ AI power users.
The Ultimate Insiders’ Guide to AI: How Founders, Investors and Executives Really Use the Technology (link).
Jeff Bezos, Nvidia, OpenAI & others invested in humanoid robot startup Figure AI at a $2B valuation.
ChatGPT for Android now has a home screen widget.