Dust is settling from Monday’s GPT-4 announcement.

Phew.

Here’s what you might have missed.

🤖 Top Stories

😬 Google says hey, we announced cool AI features too

Google picked possibly the worst day to announce their new AI features. The same day, and one hour later, GPT-4 dropped.

Let's give them their moment here.

Two big things.

They’re tempting developers with their AI carrot. Their AI language model, PaLM, is now on Google Cloud. They’ve also got a generative AI app builder.

Now, onto the stuff normal people will use.

Google's using AI to make it easier to use GSuite.

Their two examples: Gmail and Google Docs.

For Docs, they’re envisioning a future where you ask Docs for anything, and it’ll give you a rough draft. Ask it for a job description, and it’s there. You polish the details and boom. An hour saved.

Gmail is much the same - you can turn notes into a finished email. Sweet.

They even added an “I’m feeling lucky” button - if you want Gmail to rewrite it for you.

Here’s the thing. These are actually useful features!

But they're not available yet, so no one could toy with them.

Thanks that, and to GPT-4, everyone forgot.

👀 Growth over ethics, says Microsoft AI

Hyper-growth for AI means forgetting the AI ethics team.

At least, that’s what Microsoft is telling everyone. They laid off the whole team off in a recent round of layoffs.

Microsoft will of course tell you this is still a priority for them. That they have an “Office of Responsible AI”, which, while true, sounds like a Harry Potter ministry.

But the AI ethics team tackled a lot of issues that would slow down development. Issues like how adding generative AI to their Office suite might backfire catastrophically.

By firing that team, the message from Microsoft is clear - move fast and break things with AI. Default to growth. This is an arms race. Microsoft has the first-mover advantage thanks to its OpenAI partnership.

Time for trial by fire.

🤯 Intercom adds GPT-4 for AI customer support

Intercom’s got AI customer support covered, using GPT-4.

Yawn. Why is this interesting?

They didn’t feel like this was ready in GPT-3.5. Now it is.

In their blog post, they outline how the old version of the chatbot would just make things up.

So what did they do?

• Restrict information that the chatbot has access to

• Allow it to say “I don’t know” and pass the convo off to a real human

• Let it refuse to talk about things not related to your company

• Wait for GPT-4

That last one’s the real kicker.

Every AI startup needs to pay attention.

Intercom is the incumbent. In the past, startups could disrupt the incumbent with the right technology.

Companies like OpenAI have made it so easy to incorporate tech like GPT. The incumbent actually has the advantage. They already have all the data that makes GPT useful.

Simply being “Intercom + GPT” is not enough for a startup to take on Intercom.

Differentiation is crucial. Real differentiation.

GPT is breaking the innovator’s dilemma.

😎 AI tools, links, papers

1. Midjourney launches a magazine!

2. Create Doom using GPT-4.

3. GPT-3 vs GPT-4.

4. Midjourney v5 is seriously cool.

🤖 Got feedback for us?

Head over to our Twitter, @RobotRockDaily, and let us know what you’d like to see more or less of.