As promised, I’ve got a special mailbag issue this week. Thanks to everyone who sent in questions. Like last year, I picked a handful that hit some of the themes I plan to continue covering in 2025. 

I’m really concerned / worried / curious about the near-term future. Between now and 10 years from now, I think it is very clear AI will be replacing many job functions. What are we all going to do?

The leaders at the AI labs say that, yes, there will be job loss, but that doesn’t mean catastrophe. The optimistic take is that humans are creative and will invent new jobs, like they always have when technology changes things. At the moment, there’s also a macro belief among the CEOs driving a lot of the spending on infrastructure for AI that its impact will be deflationary and lead to GDP growth. 

Job displacement will still be painful, of course. Sam Altman and others believe that some form of universal basic income will be necessary to offset the economic impacts of AGI. Altman has his other startup, Tools for Humanity, already scanning eyeballs and distributing cryptocurrency. But I think it’s way too early to be seriously concerned. As Altman himself recently said, AGI is going to be declared soon and we probably won’t notice.

How much better is the reasoning on AI models, and is it actually something I should care about?

I know people who have tried ChatGPT’s o1 pro mode and notice a difference. But I haven’t seen anything mind-blowing from o1 or what Noam Shazeer at Google just put out, though perhaps I am a bit jaded by the last two years of AI hype. My advice would be to play with what you can access / afford and see for yourself.

The expense of running these cutting-edge “reasoning” models is currently keeping them at bay for a lot of people. I expect access to widen significantly in 2025. Knowing how to prompt these different kinds of models effectively remains a struggle, and I’d like to see more interface improvements in apps like ChatGPT to help teach people why they should use a reasoning model. An even better move would be to abstract away all these definitions and focus on what tools can do for people.

What kind of outlook do you see for Snap in 2025 and beyond?

Snap’s biggest problem going into 2025 is the same problem it had going into 2024: its business isn’t growing fast enough. The app itself is bigger than ever and growing quickly, but yearly revenue growth last quarter was less than Meta’s. That’s not a compelling pitch to Wall Street when you are already viewed as the underdog. Even with ads being placed in the Chat tab and the new Spotlight redesign slowly rolling out, the jury is out on if the business can rebound to the pace it needs to this year.

A depressed stock price makes it harder to recruit and retain talent, which has become more of a problem for Snap in the last couple of years. I do think the vibe could shift quickly if TikTok does end up being banned in the US or severely hamstrung by a new ownership structure. 

I continue to be skeptical of Evan Spiegel’s commitment to hardware with Spectacles. As I’ve written before, his foresight and ambition to build AR glasses is admirable. But Snap looks increasingly outgunned in hardware.

What do you expect from Meta’s glasses in 2025? 

There have been a couple of reports recently saying that Meta is planning to ship a pair of smart glasses with a heads-up display this year. I first reported this was going to happen in February 2023. Hypernova, as the product is internally referred to at Meta, will have a viewfinder for interacting with things like Meta AI and notifications.

In my write-up of the Orion prototype, I spent a lot of time on the neural wristband because it’s going to ship with Hypernova as a way to control them (while Orion’s commercial successor is still a couple years out at least). I expect this band to be the part of the glasses that surprises people the most. Using it for the first time feels like magic. As I reported in 2023, Meta is also planning a separate smartwatch as an optional upgrade with the neural capability and more features for health tracking, etc. It’s going to be a very interesting year for Meta on the hardware front.

Is TikTok going to actually be banned?

No one I’ve spoken with who is in a position to know thinks that China will let TikTok be fully divested from ByteDance. The algorithm definitely won’t be sold, but as I’ve explained before, that isn’t as important a factor as it was the last time TikTok was facing a ban.

At the same time, there is too much money and power at stake for TikTok to just disappear. President-elect Donald Trump wants to make a deal. The most likely outcome is a different version of the frankensteinian “TikTok Global” joint venture proposal that ByteDance agreed to back in 2020.

I could see Oracle staying involved this second time given Larry Ellison’s ongoing influence at Mar-a-Lago. ByteDance will most likely continue running TikTok day-to-day while divesting some of its ownership stake. The real wild card in all this, however, is Elon Musk, who has had serious TikTok envy since he bought X…

Are you more bullish or bearish on Google than you were a year ago?

Honestly, bullish. It’s going to be difficult to achieve Sundar Pichai’s 2025 mandate of making Gemini a serious rival to ChatGPT on the consumer side, but Google has a fountain of money, the technical talent, and unrivaled distribution. 

The company’s challenge is more of a cultural one. The more you have, the more you have to protect. It’s hard to get such a large, sprawling conglomerate to move fast and not care about the risk of backlash. Pichai seems well aware of this and the threats he faces, though.

Even if Google has to end its Search default payments to Apple (which I predict will be the most likely outcome of the DOJ antitrust case), doing so probably hurt Apple’s bottom line more than Google’s, as Eddy Cue himself argued last week. 

Then there’s Waymo, which may end up paying for all of Google’s “other bets” failures over the years — and then some.

What is a good book you recommend that falls in line with the things you report on?

A curse of already reading so much for my job is that I rarely want to spend time on a book. The last book I read in full was The Biggest Bluff by Maria Konnikova, which has nothing to do with tech but is super valuable if you are getting into poker. I enjoyed how her story of becoming a pro player is woven into explaining the technicalities of the game.