316
Coca-Cola's New AI-Generated Soda Flavor Falls Flat
(gizmodo.com)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 25 Sep 2023
316 points (95.1% liked)
Technology
59094 readers
3098 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
I have so many questions, none of which are answered by the article. Was the flavor really picked by an AI? If so, how did they train the AI? What kind of AI was this? What other flavors did it come up with? Did they try a bunch of them and this was the best one they could get?
This whole thing just screams marketing stunt to me, and not a particularly good one. I can't wait for this whole AI thing to just die out already. How is it that every tech fad seems to somehow end up being even dumber than the previous one (although I think the whole NFT thing might have set a new low bar)?
It’s just the latest in a long line of experimental, conceptual Coke flavours. Honestly, it’s something I’ve been saying for years; stop being constrained by imitating “real” flavours and let the flavour scientists loose, let 'em go nuts.
So far they've done:
Space (I liked that, hints of toasted carmel and raspberries)
Dream (Also good, a little bubblegummy, a little cotton candy-y, a little mangolike)
Transformation (Awful, like coke with coconut oil and a hint of turpentine)
Byte (Just decent, kind of indescribable)
Pixel (I never got to try it, it was US only, but by all descriptions it wasn't great)
Movement (A bit like theatre butter and cinnamon, it was okay but wasn't a fan)
And now AI flavour. I plan to give it a shot, but I don't expect much after their last two Tech-y flavours were eh.
You're totally right, everyone knows that blue has been one of the best flavors for a long time, yet most companies are scared because blue "isn't real".
how can blue be real if blueberries arent real
Where do you get all these Coca Cola flavours? I'm in Germany and have only ever seen vanilla, lemon, cherry and life (next to the default, light and zero).
I get them at my normal grocery store (in Canada). They're limited time, they rotate in and out, so maybe you just missed them, or maybe they're an NA thing.
I tried the XP flavored coke that is marketed towards gamers and I couldn't really tell a difference between it and Coca-Cola Classic™. Maybe it has a slightly bit more licoricish flavor? I couldn't tell because I was too busy leveling up with all of the XP I was gaining while drinking it.
Proper classic? The one with cocaine?
Mountain Dew seems to have been leaning into that, and it's mostly been good. I don't know what their latest Halloween mystery flavor is supposed to be. It's certainly nothing natural. But I like it.
Baja Blast is fucking awesome.
Dream sounds pretty fire tbh.
No, Fire TBH tasted like ashes and hose water.
I really think they should bring back Starlight, possibly outside the Coke brand. To me, it tasted like graham cracker.
"mmmm, tastes like real AI!"
I've tried most of those. The first one, which was called starlight in the states, was the best one so far. The others haven't been as good. The new AI one is probably the one I've liked the most since starlight.
I mean, that SOUNDS kinda terrible.
...just the kind of flavor that a computer would enjoy though. Hmmmmmm.........
I always try the other flavors, but can’t say I’ve liked any of them. The AI one was probably the worst of them though.
I loved Starlight and dream. Both were the kinda things that I wouldn't want every drink to be like, but wow were they cool to try out and I'd absolutely try any experimental flavours I see because of that.
I wasn't aware of most of the others though. I think these can be harder to find in some places. Most of the ones I've had were from 7-11s or similar convenience stores, but I don't usually even go to such places.
Space (I think it was branded "Starlight") was really good, Id keep getting it if it were available. Then I tried Dream (it was meh), couldn't get the Starlight I wanted, and just... ignored them from then on out. This new one is the first one Ive tried in a while; its ok, vaguely autumnal (cinnamon-y...?) but Im not hooked.
They probably trained it using data from their Coca-Cola freestyle dispensers if you’ve used one. That’s my guess.
People like Vanilla and people like Grape, so here's Vanilla-Grape by CokeAI!
Sounds horrifyingly plausible.
I'd drink that tbh
Yeah sounds like a Purple Cow, I’d be all over that
The answer is simple. They’re using blockchain NFTs to reach new market growth using AI to provide flavor solutions to consumers
I can’t wait to get Bored Ape-flavored Coke on the blockchain!
“Unsurprisingly, ‘Diarrhea Sasquatch Xtreme’ hit the mark yet failed to wow test groups,” is likely one of many test flavors removed from the article for PR reasons.
Sorry about that. I had taco bell for lunch.
Yeah, I'm guessing it was as as much an "AI" thing as everything was "i-something" about 20 years ago, or a bunch of stuff, even video game consoles, were the "something something computer" 40 years ago
The press release they link to is not especially forthcoming with information either and all they can get in terms of details is from that press release and tasting it themselves.
Companies using AI in a stupid way will die out, but the models themselves are far too useful for certain job fields (probably not yours or you wouldn't be comparing it to NFT's) for them to ever die. They're going to expand and become integrated into the data environment.
I don't know what their ai process looks like, what kind of data they trained it on, etc.
But annecdotally, I've played around a bit with chatgpt making cocktail recipes, and it's been surprisingly good at it. They sometimes need a little fine-tuning but they tend to get you in a pretty close ballpark, it's made some interesting suggestions I probably wouldn't have thought of, but nothing that turned out to be bad.
A lot of recipes tend to follow some pretty well-established ratios which means they can be broken down into some sort of mathematical formula which is something computers can actually do pretty well, and it's often just a matter for swapping out one ingredient or combination of them for another that is similarly salty/sweet/bitter/sour/umami.
For example a standard recipe for punch is 1 part sour, 2 of sweet, 3 of strong (liquor of your choice), 4 of weak (tea, juice, soda, water, etc.) and you can mix and match just about any ingredients that fit those profiles and get a drinkable punch.
I'm sure a company like coke probably has a long list of flavorings with known and well-documented flavor profiles that an ai trained on a list of proven recipes could mix and match with all day long.
That's not how a LLM like Chatgpt works. It's not referencing cocktail recipes, compiling their ingredients, seeing commonalities, figuring out mathematical formulas, and then experimenting with the variables. That kind of thing is still probably a decade or more away, if not decades.
If you're curious, see what people have tried to do with AI generated recipes that fit nutrition guidelines and how they never add up correctly.
From what I've seen of cocktails and recipe books, there's probably a lot more of them than you realize. I guarantee you there are thousands of cocktail recipes you've never heard of that have been written into published recipe books.
All of that to say that Chatgpt is basically just making up arrangements of words that based on its training should go together.
Yeah I've seen people recommend ChatGPT for meal planning and recipes, and it's mostly fine for common, simple recipes but it does super weird things when you ask it for something nonstandard, that has a lot of variations, or with dietary restrictions. Like it repeatedly gave me recipes with my allergens with a note to check package for said allergen and other weird things like claiming frozen vegetables take 10 minutes to roast in the oven. It's useful for certain things but it's not really intelligent.
AI isn't going to die out but hopefully it'll get quieter and more boring/focused.
It will absolutely die out in that it will go back to what it was previously. We've been using "AI" for decades now, only it's better known under the name machine learning. This latest surge in interest is just a bunch of marketing hype and a bunch of executives too stupid to realize they're being fed a line of bullshit by contractors promising if they hire them to make an "AI" they'll be able to fire their entire workforce and dump their salaries into fat executive bonuses. Just like all the previous tech fads this will stop being the hot thing once enough of these douche-bags get burned and even the dumbest of them learns that no, you can't just replace your entire workforce with "AI" and call it a day.
The tools now are far better. You can slap together something useful with some basic Python knowledge. Hardest part is mixing up the data into giving you good results.
It'll hang around, but I don't think Nvidea's market cap is justified. It'll crash hard, but it could be tomorrow or three years from now.
I don't think it's going to end up impacting most industries all that much. Low end call centers will probably be impacted, but they were already being displaced by automated call trees and such even before this latest fad. At some point trucking will be displaced by self driving trucks, but that tech is looking to be further off than it initially seemed as well thanks in part to some pretty high profile accidents and renewed scrutiny from various governments. Beyond that the impact in other industries is looking to be fairly minimal. You'll see smarter tools being rolled out to let people do the things they were already doing faster, but just like the "magic clone" tool in Photoshop while it will make some tedious time consuming activities much faster, it won't really fundamentally change things.
Honestly the biggest impact is most likely to be on crime, with these various tools being leveraged by criminals to make increasingly convincing scams, phishing attacks, and even worse things.
In biology we've been using machine learning for a long time now so the AI super hype out there is pretty funny to me. It's for sure useful with stuff like predicting protein folding and analyzing genes and stuff, but it's all hyper-specific stuff just like it has its always been. Good for removing tedium for sure as its the reason we can even know the human genome because it would take literally forever to sequence it without modern tech, which we did in the in the 90s and finished in 2003.
My big hope is that all this hype will get people to invest in proteonomic technology which is 100% a great use case for AI and also the future.
Well, on the label of the ones I tried it said co-developed by AI.
So yeah, probably marketing stunt
That said, if it hadn't been artificially sweetened, I would probably have preferred it to the normal one. Felt like it had more flavor. Similar to Fritz Cola from Germany.
Deep Learning at least can produce useful tools here or there. No one has yet to come up with a good idea for why NFTs should be a thing. Though I'm sure someone will come along with their niche use that, on further consideration, doesn't actually solve anything after all.