linkedin newsletter

Gemini Season

Frederic Lardinois

09 Feb 2024 — 6 min read

Generated with AI by Microsoft Designer

The last few months have been rough for the news business. I was going to write something about that last week, but then it was TechCrunch's time to get hit with a round of layoffs (my team and I got spared, thankfully). The company also shut down our subscription product, TechCrunch+ (a.k.a. Extra Crunch).

There isn't much more I can say about the details in public, but it's no secret that the entire industry is struggling. In tech media, you can add the overall pullback in marketing spend to that, given how dependent most sites are on ads from major tech brands. It's good to have a conference business.

I think in the long run, niche sites will be just fine. I worry more about general news sites. The Messenger clearly couldn't make it, despite getting traffic (though it surely had plenty of other problems, too). But traffic isn't what it used to be when it comes to monetizing a site. Sidenote: I finally read Ben Smith's Traffic, so this has been top of mind for me, even as I think the book is a bit too myopic in its focus.

News brands come and go. I've been lucky to have a very stable career so far, but I'm also very aware how unusual that is in this business. For the individual journalist, the publication you work for is often everything. It's a paycheck, for sure, but there's also the umbrella of that publication's reputation.

Unless you have an established brand, you can't just go and create a newsletter and charge $10 a month for that – and not everybody can build that brand anyway. Plus, not every topic lends itself to the newsletter format and even if it does, it's going to take a very long time to build up that business. Not everyone gets a 27,000 subscriber list from their former employer as a parting gift.

Enough about that for now, though. I'm sure I'll have more to say about it in the future.

Here is what I wrote about this week, with a bit of extra color for good measure:

Google Gemini Ultra

Google finally launched its best large language model to the public this week. I don't have a good sense of how well this landed for Google, but having used it for a day now, Gemini Ultra 1.0 does feel like it's on par with GPT-4. That's the expectation Google had set when it launched Gemini Pro and it lives up to that.

Still, it's a bit uneven at times – sometimes it'll happily summarize my inbox, for example, other times it'll tell me that it doesn't have access to it – but it's exactly this integration with the rest of the Google products that gives it an edge, at least in the consumer and SMB space.

Gemini always had a bit of a tendency to hallucinate and I've found that with Ultra as well.

I'm also glad that Google retired the old Bard and Duet AI brands in favor of just 'Gemini.' That branding was always confusing and Bard didn't have the best of starts, so a fresh name may just help here. What is still confusing, though, is that you need a Google One subscription to access Gemini Ultra, but that also helped Google push its One business to over 100 million subscribers now.

On Android, you can also have Gemini replace the Google Assistant. What does that mean for the future of the Google Assistant? Google won't say, but the writing is on the wall.

AWS and Amazon remain nowhere to be seen in this business (and yes, I know they have the Titan model, but be honest, until you read this sentence, you had forgotten about that, too).

New Mozilla CEO

I didn't write about this one, but I do have some thoughts:

Mozilla has been looking for a second act for a very long time. Firefox – and the Google search deal – keep it afloat, but Firefox is dwindling in its relevance. I would like Mozilla to focus on that, but instead, the organization (with its complicated relationship between the Foundation and the commercial side) is more focused on other projects that have far less defined of a mission. And yes, people want a browser that respects their privacy, but clearly, that isn't enough of a selling point.

When I talked to Mozilla president and executive director Mark Surma earlier this year, we talked a lot about the organization's plans around AI – and specifically the responsible use of it. I think Mozilla has a chance to play a role here, but it hasn't really established itself as a player here (and it doesn't have the money to build its own models that could compete with the open models of Meta, for example).

I don't know anything about the new interim CEO Laura Chambers. The announcement was big on corporate speak (" Outstanding Execution: Focus, Processes, Capabilities") but low on specifics.

All of this doesn't make me feel more optimistic about the future of Mozilla.

Also from me in the last two weeks: