This is my concern as well. How successful is it in selecting the correct tool out of hundreds or thousands?
Different to what this integration is pushing, the LLMs usage in production based products where high accuracy is a requirement (99%), you have to give a very limited tool set to get any degree of success.
There is a generalized reasoning that current LLMs still miss. It's hard to put a finger on it. Things like hallucinations show that there isn't a self awareness of thought. "Thinking" models are getting closer.
From most of my network trying to make products based on LLMs, excluding cost, the biggest hurdles are hallucinations, and seemingly "non-sensical" reasoning or communicating. Subtle choices that just "feel" not quite right. Particularly when the LLM is being constrained for some activity.
Open-ended chat doesn't show these flaws as often.
An interesting side effect of this low-level bit mapping is that various banks authorization logics can be manipulated to increase auth rates by subtle bit flipping across various fields.
All the big fintech companies have ML running over changes to identify what results in the highest auth rates on a per bin basis.
I would be very surprised if bit flipping and ML were really used here, do you have any source?
While for sure there's a lot of signal and value in monitoring auth rates per BIN per payload, flipping bits can be extremely disruptive and counterproductive. From doing the wrong operation to being fined by the schemes, it's a lot of risk for not a lot of gain when these fields can be tuned ad-hoc for the few card issuers that deviate from the standard/norm.
To define "me" in this context basically requires solving the Hard Problem of Consciousness [0].
To an external observer, all life, including humans, are nothing more than a complex set of electrical and chemical reactions, and in theory, with sufficient technology, you could take a perfect subatomic snapshot of someone's brain and predict exactly how they would react to any stimuli.
But then there are the first-person perspective experiences. We have words for it like "consciousness" or "mind", but no scientific explanation on how it works and why it shuts off when we go to sleep. The religious will explain it with the concept of a soul.
I don't think we can ever solve this Hard Problem. But I can claim to believe that if someone did snapshot my brain and recreate a perfect living replica in a new body, that I would not experience what that replica experiences. It might have a completely new consciousness behind it, or it could be a P-Zombie [1].
Maybe I'm actually the only conscious being in the entire universe and everyone else is a P-Zombie. Somehow, to me, that makes more sense than consciousness appearing out of nowhere.
> Maybe I'm actually the only conscious being in the entire universe and everyone else is a P-Zombie. Somehow, to me, that makes more sense than consciousness appearing out of nowhere.
That's because you live in Plato's cave, we all do.
Not necessarily. Some of us may have torn the illusory veils of Maya, and gleefully surf the waves of Indra's ocean from island to island, tasting all the delciously different somas.
I have been trying to use these jailbreaks to generate a new prompt for future jailbreaks but I consistently hit warnings on openAI's content policies.
The games themselves but particularly discord as well. Discord is the modern gaming social network. Talk and hangout with friends across multiple games.
This is what I am hoping for. I want to have my tablet mode with iOS apps and interface. Then I want to switch into laptop/desktop mode, being able to spin up any desktop application.
Even more a dream state: would want a docking station that allows me to use my multi monitor setup.
I worked for a retail analytics company that would do things like track WiFi/Bluetooth devices, faces, and general shopper traffic.
It had been a few years since I've worked there but the real problem was that all of this didn't work. All these promises to our clients that's you would get data equivalent of online shoppers, in store, but it simply wouldn't work. Guess people age, gender, previously seem face, tracking where they walked. All horribly inaccurate.
Though apparently it's an industry wide thing as "imputing" was common practice. Don't have the data? Make it up!
I don't know about retail but if you are ever in Las Vegas or Reno, go into a casino and walk past every slot machine to the reservation desk and ask for a room. Availability and prices will be far worse than booking after gambling or even phoning from the parking lot.
Isn't this just price segmentation by contact method? If im already at your hotel asking about rooms I'm likely less price sensitive than someone calling in advance. But If i give you my players card likely get some kind of paltry discount.
Yeah, well -- LV is the original source of much of what we know today about data mining. So, it makes sense that they would still be decades ahead of the rest of the field.
Yea you just “stitch” the data together to get your perfect customer data profile.
Actually you end up with about 5% of your customer base properly tracked, and those people are already buying the most stuff anyway. I would bet that nailing the supply chain and focusing on your margin would be a better investment in most cases.
True. Most of it is marketing BS. Overpromise is the norm. But if you read the comments here, you see that even technical people are buying that. And possibly it will work later, so we can worry now.
But there's no reason that this couldn't work well with improvements in sensors and implementations. If you don't think this is possible in a highly reliable way, you are wrong.
Technology is ever advancing and improving and I’m sure eventually we will get there. Or we’re already there, like I said, I’m out of the loop for past few years so maybe some amazing improvements have happened.
Another inhibitor of this technology is how cheap/behind retail companies tend to be.
We want advanced analytics.
Ok, we can set up these cameras and the like which will send everything to our servers for computation.
We only have dsl at the store.
Ok, then we need to set up a server(s) at the store.
We only want to spend 2k max per store, and use sub-sub-sub contractors to actually set up this equipment.
All of this pain and trouble scared me away from working with retail focused companies for past few years.
This. I am a lead developer in a leading FR company, and this comment right here hits the nail on the head. Every one of your points is an active issue with retail. It's a pain in the asshole.
With optical cameras, wifi and Bluetooth signatures, location data, 3rd party databases, and ML, there is more than enough data to get an ID enough shoppers to care. None of that tech is prohibitively expensive. It just requires someone to build out a cost-effective implementation with off the shelf parts.
There's a lot more "just" to deploying anything of the sort to a 1,000+ physical store footprint.
At minimum, what's the local network situation? Where are you storing and processing the take? Are you installing edge compute or transferring it out? If the later, what are your latency requirements?
I try and take a healthy dose of "maybe it's a harder problem than I thought" whenever something is valuable, looks easy, yet no one is doing it.
This plan all falls apart when the retail CEO has "his nephew manage the project, 'cause he's good with computers". Then a 20 year old more concerned with girls and video games appears. He has zero technical experience beyond video games, and attitude that you're his flunky. Seen it. Multiple times.
Generally, Adyen is cheaper than Stripe but Stripe is easier to integrate. When considering moving from Stripe to Adyen, It comes down to a equation of dev cost/resources to integrate to Adyen compared to the amount of reduced fees/increase auth rates.
The more volume you process, the more a switch makes sense. and this difference of mentality is reflected by the fact Adyen has minimum monthly invoices when Stripe does not.
Different to what this integration is pushing, the LLMs usage in production based products where high accuracy is a requirement (99%), you have to give a very limited tool set to get any degree of success.