Deep breath. There’s no sense in trying to outcompete Google in burning cash. They’ve got time to wait until there’s the beginning of commodification of the tech, and a large profitable market to be had.
Sometimes company’s just don’t do good enough.
It remains to be seen whether this was a smart move, or just flailing money at the wall
Zuck tried and flailed with the metaverse. That was a huge waste, but he can afford it and fortune favours the brave.
Not everyone has to make the same move at the same time.
If those don’t seem like right or good moves, I can’t imagine much will impress you in this world.
This is actually one of the hardest frontier problems. The "general purpose" assistant is one of the singular hardest technical problems with LLMs (or any kind of NLP).
I think people are easily snowed by LLMs' apparent linguistic fluency that they impute that to capability. This cannot be further from the truth.
In reality a LLM presented with a vast array of tools has extremely poor reliability, so if you want a thing that can order delivery and remember your shopping list and remind you of your flight and play music you're radically exceeding the capabilities of current models. There's a reason successful (anything that isn't demoware/vaporware) uses of agentic LLMs tend to narrow-domain use cases.
There's a reason Google hasn't done it either, and indeed nor has anyone else: neither Anthropic nor OpenAI have a general purpose assistant (defined as being able to execute an indefinite number of arbitrary tools to do things for you, as opposed to merely converse with you).
For the purposes of the exercise, let's conservatively say, maybe ~2000 tools covering ~100 major verticals of use cases. Even that may be too narrow for a true general purpose assistant, but it's at least a good start. You can slice the sub-agents however you'd like.
If you can get recall, for real user utterances (not contrived eval utterances authored by your devs and MLEs), over 70% across all the verticals/use cases/tool uses, I'd be extremely impressed. Heck, my thoughts on this won't matter - if you can get the recall for such a system over the bar you'd have cracked something nobody else has and should actively try to sell it to Google for nine figures.
So much attention, effort, and tooling has focused on getting llms better at writing more and more code. They can grep and curl and run scripts and iterate and build things really fast, and maybe even maintain it if given enough guardrails and direction.
But it turns out we have had a _ton_ of useful training data for models to work with for software. Not just books or docs, but examples, tests, snippets and full programs for just about any language. Show me a stackoverflow with playwright scripts or API calls (hah, as if thats possible) to build itineraries from delta, aa, united, priceline, expedia, etc, .... which is one part of one piece of the ai assistant pipe-dream.
I don't think its impossible as these tools get much smarter and more generally capable that we get decent assistants in other constrained, non-software domains, but it will take very good companies focusing on it for a long time. Much like any product that try to do these sorts of things.
Its so easy for programmers in our bubble to overlook the complexity involved in automating or even _describing_ simple tasks that humans navigate everyday via habit, learning, experience, and perception...all things that llms struggle with constantly.
There’s not just one specific solution to it either, there’s a whole class of tooling for it. And I doubt google would pay 9 figures for something that’s built on top of libraries they put out using models they developed.
As of August 1st ‘we’ (as in, I personally developed with my company, and have been paid for with real dollars which are now sitting in my bank) have a F100 using this tech in production.
As for the no true Scotsman fallacy you’re putting in front of yourself, I will let you deal with that but I would like to see how you came up with the maths.
This does seem like an embarrassing fail, but even Google has not completed replacing Assistant with Gemini. There have also been lost functionality (maybe temporary) in the process.
Every business has to make tradeoffs, it's just hard to imagine that any of these decisions were truly worthwhile with the benefit of hindsight. After the botched launch of Vision Pro, Apple has to prove their worth to the wider consumer market again.
Doesn't somebody (not named Nvidia) need to make a serious profit on AI before we can say that Tim Cook failed?
OpenAI and Anthropic aren't anywhere close. Meta? Google? The only one I can think of might be Microsoft but they still refuse to break out AI revenue and expenses in the earnings reports. That isn't a good sign.
I won't pretend to know exactly how the AI landscape will look in the future, but at this point it's pretty clear that there's going to be massive revenue going to the sector, and Moore's law will continue to crank.
I see what you're saying though. In particular is first generation gigs data centers might be black holes of an investment, considering in the not too distant future AI compute will be fully commoditized and 10x cheaper.
"failed to skate where the puck was headed" assumes that we know where the puck is going to be. We don't.
Everyone is skating towards that same spot while Apple is over by the blue line practicing their swizzles. They sure look like they're doomed. But large groups of people have skated to the "wrong spot" thousands of times. That's the entire point that Gretzky was making with his quote. He's not big enough, strong enough, fast enough to get in that scrum. They're all fighting it out and the puck slides away. To him. All alone.
Maybe that is Apple, maybe it's not. I mean, they're still learning to skate while everyone else is playing hockey.
The Mac is something like 30 billion in revenue per year, and 10 billion in profit.
The entire "generative AI" "industry" is struggling to reach 30 billion in revenue even with their creative accounting (my free Perplexity that comes with Revolut is somehow counted at full price, even though I never paid anything, and I'm sure Revolut doesn't pay full price), and gross profit is deep in the negative.