(June 2024) AI Drive Thru's Failed But The Rollout Continued

The have solid intel to suggest the girls took the burgers.

I don't believe it was ai's fault at all. I believe they hit that girls school on purpose to kill the kids of a bunch of their leaders and elites.

That sort of thing is right up Israel's alley, killing the families of people they don't like. Our president endorsed the same thing at one point campaigning too.

If only they could cool data centers with coca cola /s

Ain't gotta be slow to need a sarcasm tag, but thanks regardless!

Remember when they tried outsourcing that to India?

The more I sit on it, the more enraged I am about the news.

They most likely stole the result. They dropped a massive proof with no human understanding it (come on, how many can understand math at that level? 5 world wide? And definitely none at OpenAI), they dropped hundreds of thousands of line of validation (see above), burned a significant fraction of the metaphorical Amazon in the process, dumped more money than most math departments yearly budgets, and claimed victory.

Technically a team was already working on it and they lied by making A.I look better. That's just a marketing stunt it didn't do shit.

Add to that, we have found the Altman, but we are still in danger until we discover the whereabouts of the Controlman and the Deleteman.

He should embrace it.

"Hello. I'm Jake your burger....... assistant. I'm here to take your order...... I'm here with you until the end. Let's do this. Would you like..... fries.... or a large refreshing soft drink.... with your order?"

I mean what can you say that doesn't risk getting you fired?

I'm sure they're putting all those costs onto the franchisees. McDonalds the corporation won't feel a thing.

This is why McDoubles are like $3 instead of $1.

McSchadenfreude

I stopped when they started charging sit down restaurant prices for food barely higher quality than what prisoners eat.

AI saving lives by failing to sell McRibs was not on my bingo sheet

+1, I also stopped shopping from any market that forces self-checkouts. I waste more time, I scan my own stuff, and for what exactly? So the company can earn more money by kicking out their employees? Why the fuck don't we even get a discount for "working briefly" in their market? Nothing these companies pull off ever benefits the society.

Offering the option is one thing, forcing it down your throat is another.

I bet some amount of that was deliberate. Kinda like how they started charging extra for value items "dressed like a mac" when people realized the only real difference between a McDouble dressed like a mac was the central bun and it was much cheaper. So they started charging most of that price difference for a squirt of sauce that replaced a squirt of ketchup because McDoubles were for people too poor to afford a Big Mac, not for people who could afford a Big Mac but thought it was a rip off. At least that's what I remember from the time before I realized McDonald's wasn't ever worth going to. I can't think of a single fast food place with burger patties as pathetic as McDonald's (though I have a feeling the US has other chains competing for that spot).

When you're dealing with a human, they can make customizations that might upset the greedy little piggies at the top that would rather gouge people for any improvements and hate when employees give out freebies even if they cost nothing and generate good will for the business. Hell that might have been the difference between then and now for my willingness to go there at all, as it felt less human and more corporate hellish when they started charging for it.

There are a lot of salaried people out there that also wouldn't be able to get my order right in the drive thru

What annoys me is that automating order taking doesn't require a full LLM. They could easily program a voice to text system with a library of food items, basic voice commands, and have the outputs display on the screen so that the user can verify and correct the order verbally.

Like smart tech doesn't need to know anything beyond it's job. But they're trying to normalize these massively superflous systems because the economics of big corporations has completely decoupled from reality. && Fascism.

Naw, audio doesn't take much, definitely not as much as a full time employee. The text models probably don't need to be very large for this, the whole thing can probably run on a modern phone.

With all the optimization that has been done, everything can be done pretty efficiently except for video and good image gen.

The main issue with all this is probably audio quality (wind + car running + garbage microphones)

Inference is actually relatively cheap. A company the size of McDonald's could also negotiate a deal.

Tokens are expensive for you but for the labs tokens are cheap, training is expensive.

Get your law degree at Costco now!

I have very limited experience in LARGE corporations. But it seems to me those are all "putting out fires" positions requiring some level of rapport with key individuals and usually privileged knowledge.

Not saying it justifies the exorbitant salaries they get. Just explaining why in a "shareholder profits first" system we're not seeing AI's running corporations.

Especially given the results of AI run vending machines / stores.

Even if AGI was "just around the corner", it'll be built on the bones of the broken-ass LLM situation we have now. The optimism doesn't just cover our ability to iterate towards it, it extends to the ability of said AGI's ability to unlearn bluffing, gaslighting, and just fabricating things. I'm super doubtful about all of it, especially the latter part.

Honestly, human communication is insanely complex, we just happen to be heavily optimized for it.

I mean, ultimately it would be slower than just having a person. Serving less customers per hour is part of it, but also people may go elsewhere if they see a lineup.

Just speaking as someone who worked drive thru at mcdicks for a year. Not saying I endorse it.

IKR? The staff at rage rooms keep insisting I'm only allowed to hit the inanimate objects. SMH.

There's so many models too, maybe they cheaped out and wasn't viable for the more expensive ones

The last time I went to Wendy's, the voice recognition and model were both surprisingly good. I had a "natural" conversation when it told me the crispy chicken BLT wasn't on the menu anymore and it acknowledged my annoyance that a computer was offering to add a slice of bacon for $2.50.

That said, the very existence of this new drive through is why I havent gone back. A shame because the 4 for $4 got me through some hard times

Frankly, I doubt the problem was even the software at all; I think it was doomed to failure simply by shitty speakers and microphones.

As soon as they can build a supply chain for fully automated kiosks I imagine.

Carl's Jr. Fuck You, I'm Eating

For a company that loves power its crazy to think they are just ok handing over a decent chunk of that power to another company

While at the same time fighting unions every step of the way.

Doing significantly worse than McDonalds humans is pretty bad though.

That's an interesting point. AI has a current limitation of not knowing that it doesn't know something.

It's not a fallacious statement. You're correct that it would still not be an appropriate tool, but that statement doesn't claim it would be.

Thru was a legit spelling of thurh before through came along, and although it's now considered informal, it's commonly used on road signs because it's compact, and "drive-thru" is considered a legitimate use. Doesn't confuse meaning or pronunciation, either.

Now, the "greengrocer's apostrophe" being used for a plural is a perfectly reasonable gripe. Can easily confuse meaning, and wasteful of a space/character!

Wait how is ‘thru’ a word-like jumble of letters but ‘through’ is fine?

For someone so interested in language, it’s baffling to me that you don’t seem to understand how language works. People have been laughing at stupid prescriptivists since words have been written down.

Or do you just have a humiliation fetish?

Pointing out that there were major pilot projects in 2024 by major companies, whose results indicated the 2025 mass rollout of AI into everything would fail, is not making it easier for tech bros to dismiss opposition they were already ignoring.

They shouldn't be using a general LLM. They should be training a model explicitly for fast food restaurants

Or perhaps they should pay a proper wage to human employees who can do a better job

They still can't reason, because that's not how they work.

It's glorified auto-complete that cannot tell the difference between training data and input commands.

Albuquerque New Mexico.

Stop drinking the kool-aid my man, just because the AI companies label things as "thinking" and "reasoning" doesn't it's the same thing as those terms when applied to humans. There is an argument to be made the analogy works well enough with some terms, but especially with reasoning it's pure marketing BS.

Internally reasoning is called "internal recursion" and it's more of a scratch pad. It allows the model to split up tasks into smaller bits, where it would choke if it tried to tackle the whole thing at once. It goes through the question, tries to come up with results and writes those down in its scratch pad. That whole scratch pad including everything that came before (system prompt, prompt, context etc) gets put into the model again to produce the final result.

Not to be confused with Chain-of-Thought, which is similar but different. That's where multiple passes get made and the tasks are actively broken up with different system prompts. This especially helps in tasks where the context would run out, but doing the same thing in smaller steps does work.

It's very annoying how AI companies heavily anthropomorphize everything related to LLMs, right down to calling it AI instead of just an LLM and leave the intelligence out of it. This easily dupes and confuses people into thinking the tool is something it's not. People think because they use terms like reasoning and thinking that these models have motivations, internal dialog, actual thoughts and even do things like planning, anticipating and predicting. This simply isn't the case, it's a mathematical model, you input numbers and it spits out numbers. When you input the same numbers, the same numbers come out (although parts can apply randomization, usually controlled through a parameter called temperature). It isn't alive, it isn't thinking or reasoning or smart. It doesn't have motivations or evil intentions. It's just numbers.