WrapPRO is free this week. See the inside scoops, expert analysis and exclusive data subscribers get daily. Click here for more information.
Box CEO Aaron Levie has an uncanny ability to explain complex tech moves in simple language. Whether it’s Meta’s decision to open source an AI model or Apple’s plan to build its own “Apple GPT,” Levie gets to the bottom of issues fast. And so, in a week filled with major news, it was time to get on the line.
This week, I spent an hour with Levie — recording a special edition of “Big Technology Podcast” that’s releasing Friday — and we went deep into the tech giants’ recent AI power plays, discussing their moves and tackling broader questions around jobs and creativity.
Below are the highlights from our conversation, edited lightly for clarity.
Why some believe Meta open sourcing its large language model, Llama 2, is meant to harm competitors:
If you can reduce the proprietary nature of your competition’s technology stack and commoditize that for the market, on the margin, you might be able to reduce some of the value proposition, or some of the competitiveness, your competitor might have. It’s probably a little bit of a stretch in AI right now for Meta, since they ultimately just want more usage of their platforms, and AI can be a contributor to that. So, I think the take is probably, if they can commoditize something Google is otherwise going to make proprietary, or charge for, there’s some kind of game theory in that.
Meta’s most likely reason to open source the Llama 2:
Open sourcing brings in great talent. It puts Meta at the top of the food chain in terms of leading technology companies. It has everybody studying their technology and their models. They become more of a standard-bearer for the industry. That all generally accrues good things.
Whether people actually want chatbots, no matter how good the AI performs:
The challenge is you’re always competing with a powerful graphical user interface where in just two clicks, without any verbose typing, you can get exactly that same answer. I think the jury’s out, to be totally honest, on what the form factor ultimately will be for a lot of these AI interactions.
The clear enterprise application for generative AI:
In a lot of enterprise use cases, you’re not competing with a graphical user interface, you’re solving a problem that could not be solved previously. In our business, you could not go to a dataset of information made up by documents, ask that dataset a question and have an AI model actually answer that question. It just was not possible. This was not something search could even do. Now, these large language models, combined with this idea of a vector database, can solve that problem.
Where Apple might compete with its rumored “AppleGPT” project:
Their play should should be pure utilitarian. It shouldn’t be a social thing. You should be able to grab your phone and just say, “Hey, can you order me a pizza,” and then a little dialog comes up, and it says, “Hey, here’s where we’re going to order the pizza from, click Yes.” That seems like the most obvious thing on the planet that Apple should just do. Or, “‘”Hey, please write down a note for me to go do this thing, and remind me later,” and it just gets it right every single time. Obviously, that’s just like Siri-plus-plus. But I think Apple has both the device and the software prowess for it.
Maybe they were late to large language models in this current ilk by nine months or something, but the good news is that everybody’s still buying their phones. And I think people would switch overnight, if there was an integrated application right on the device. If they put their mind to it, and the use case was a smart assistant on your phone that could do everything, I would be very optimistic.
Why AI companies struggle to build moats:
It seems pretty tough to be an AI model trainer, unless you are Meta, Google, Microsoft, OpenAI or Anthropic. And you definitely don’t want to be one of these pure thin interfaces on top. You’ve got to really establish a high degree of value, and probably workflow, and a data moat associated with your product.
Microsoft vs. Google:
Any midsize enterprise on up will not make the switch on the basis of AI pricing. So if you’re a 5,000-person company, and you’re already on Exchange, and you’re already on Office and you’re already on Teams, AI will not be the determining factor for switching out your entire infrastructure. Microsoft knows that, and they’re in a good position to benefit from the stickiness of their platform.
Is ChatGPT getting dumber?
The core novelty is now over. Now it’s actually, “Are you telling me very useful, reliable information?” Our level of criticality has just gone up, because there’s no novelty factor overwhelming or outweighing the critical component that we wouldn’t have had nine months ago.
A practical view of AI replacing jobs:
Right now, these things can do one discrete information-oriented task, basically, at a time, before they need a human to review what they’ve done, and then move to the next thing. There’s just not that many jobs relegated to only one kind of thing. Even in the scenarios where people have tried to say “The paralegal job could be automated,” it’s just simply not true.
These job functions are pulling together too many functions across email, and a community, and another communication tool, and a manager, and going into some other system, and then reviewing something, and then writing something. We’re just nowhere close for the AI to successfully wrap all those scenarios together. The cost of an error is so high that no one is willing to put that liability on the line.
Will AI harm or help creativity
If you’re a director, or a cinematographer, and you have to figure out what shot you want to do for some film, you spend X amount of time thinking that through and storyboarding that. The capability you now have is you can look at 1,000 simulations and pick out exactly the kind of shot you want to go with. Now, your creativity — combined with the automation benefit of seeing lots of options and testing different scenarios — will actually just get better. Something that would have taken months, or quarters, is now shrunk into a much shorter period. We imagined AI is gonna go replace those types of jobs. It won’t, it will be another tool in their arsenal.