this post was submitted on 19 Nov 2023
497 points (86.9% liked)
Technology
60071 readers
3571 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 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Probably as much as I care about most other people's thoughts on AI. As someone that works in AI, 99% of the people making noise about it know fuck all about it, and are probably just as qualified as Barack Obama to have an opinion on it.
What do you do exactly in AI? I'm a software engineer interested in getting involved.
I work for Amazon as a software engineer, and primarily work on a mixture of LLM's and compositional models. I work mostly with scientists and legal entities to ensure that we are able to reduce our footprint of invalid data (i.e. anything that includes deleted customer data, anything that is blocked online, things that are blocked in specific countries, etc). It's basically data prep for training and evaluation, alongside in-model validation for specific patterns that indicate a model contains data it shouldn't have (and then releasing a model that doesn't have that data within a tight ETA).
It can be interesting at times, but the genuinely interesting work seems to happen on the science side of things. They do some cool stuff, but have their own battles to fight.
That sounds cool, I've had roles that were heavy on data cleansing, although never on something so interesting. What languages / frameworks are used for transforming the data, I understand if you can't go into too much detail.
I did wonder how much software engineers contribute in the field, it's the scientists doing the really interesting stuff when it comes to AI? Not surprisingly I guess 😂
I'm a full stack engineer, I was thinking of getting into contracting, now I'm not so sure, I don't know enough about AI's potential coding capabilities to know whether I should be concerned about job security in the short, or long term.
Getting involved in AI in some capacity seems like a smart move though...
We do a lot of orchestration of closed environments, so that we can access critical data without worry of leaks. We use Spark and Scala for most of our applications, with step functions and custom EC2 instances to host our environments. This way, we build verticals that can scale with the amount of data we process.
If I'm perfectly honest, I don't know how smart a move it is, considering our org just went through layoffs. We're popular right now, but who knows how long for.
It can be interesting at times, but to be honest if I were really interested in it, I would go back and get my PhD so I could actually contribute. Sometimes, it feels like SWE's are support roles, and science managers only really care that we are unblocking scientists from their work. They rarely give a shit if we release anything cool.
Tbf everyone is entitled to have an opinion, including Obama
There is a tad bit of difference between caring about an opinion and tolerating one. Obama's opinions on AI are unqualified pop culture nonsense. They wouldn't be relevant in an actual discussion that would cite relevant technical, economical and philosophical aspects of AI as points.
Sure, care about it or don't, I don't care. It was the "being qualified to have an opinion" bit I didn't like. I don't have to qualify to have an opinion and I can write an opinion piece and sure enough, less people will read it than Obama's. I might not be qualified to teach on that subject but everyone is qualified to build one's own opinion.
But maybe that's just overly pedantic on my side. You are qualified to have a different opinion.
I know this was once said about the automobile, but I am confident in the knowledge that AI is just a passing fad
Why? It's a tool like any other, and we're unlikely to stop using it.
Right now there's a lot of hype because some tech that made a marked impact of consumers was developed, and that's likely to ease off a bit, but the actual AI and machine learning technology has been a thing for years before that hype, and will continue after the hype.
Much like voice driven digital assistants, it's unlikely to redefine how we interact with technology, but every other way I set a short timer has been obsoleted at this point, and I'm betting that auto complete having insight into what your writing will just be the norm going forward.
It's just a Chinese room dude, it doesn't actually do anything useful
The Chinese room argument doesn't have anything to do with usefulness. Its about whether or not a computer that passes the turing test is conscious. Besides, the argument is a ridiculous one to begin with. It assumes that if a subcomponent of a system (ie the human) lacks "understanding", then the system itself (the human + the room + the program) lacks understanding.
Anything else aside, I wouldn't be so critical of the thought experiment. It's from 1980 and was intended as an argument against the thought that symbolic manipulation is all that's required for a computer to have understanding of language.
It being a thought experiment that examines where understanding originates in a system that's been given serious reply and discussion for 43 years makes me feel like it's not ridiculous.
https://plato.stanford.edu/entries/chinese-room/#LargPhilIssu
You not having a job where you work at a level to see how useful AI is just means you don't have a terribly important job.
What an brain drained asshole take to have. But I've seen your name before in my replies and it makes sense that you'd have it.
AI is useful for filling out quarterly goal statements at my job, and boy are those terribly important... 😆
What?
At best you're arguing that because it's not conscious it's not useful, which.... No.
My car isn't conscious and it's perfectly useful.
A system that can analyze patterns and either identify instances of the pattern or extrapolate on the pattern is extremely useful. It's the "hard but boring" part of a lot of human endeavors.
We're gonna see it wane as a key marketing point at some point, but it's been in use for years and it's gonna keep being in use for a while.
I agree with most of what you're saying here, but just wanted to add that another really hard part of a lot of human endeavors is actual prediction, which none of these things (despite their names) actually do.
These technologies are fine for figuring out that you often buy avocados when you buy tortillas, but they were utter shit at predicting anything about, for instance, pandemic supply chains....and I think that's at least partially because they expect (given the input data and the techniques that drive them) the future to be very similar to the past. Which holds ok, until it very much doesn't anymore.
I’m sorry, they aren’t good at predicting?
My man, do you have any idea how modern meteorology works?
A ton of data gets dumped into a ton of different systems. That data gets analyzed against a bunch of different models to predict forecasts The median of al those models is essentially what makes it into the forecast on the news.
Well, I would disagree that they don't predict things. That's entirely what LLMs and such are.
Making predictions about global supply chains isn't the "hard but boring" type of problem I was talking about.
Circling a defect, putting log messages under the right label, or things like that is what it's suited for.
Nothing is good at predicting global supply chain issues. It's unreasonable to expect AI to be good at it when I is also shit at it.
They make probabilistic predictions. Which are ok if you're doing simple forecasting or bucketing based upon historical data, and correlates and all of that.
What they are crappier about is things that are somewhat intuitively obvious but can't be forecasted on the basis of historical trends. So, like new and emerging trends or things like panic buying behavior making it so the whole world is somehow out of TP for a time.
I'd argue that relying solely on "predictive analytics" and just in time supply chains aggravated a lot of issues during the big COVID crunches, and also makes your supply chain more brittle in general.
All predictions are probabilistic.
AI indeed isn't great at modeling complex or difficult to quantify phenomenon, but neither are people.
Our recent logistical issues are much more based on the frailty of just in time supplying than the methods we use to gauge demand. Most of those methods aren't what would typically be called AI, since the system isn't learning so much as it's drawing a line on a graph.
We didn't actually run out of toilet paper, people just thought we did and so would buy all of it if they saw it in the shelves. It's a relatively local good, so it didn't usually get caught up in the issues with shipping getting bogged down, it's just that people chose to override the model that said that stores should buy five trucks full of TP because it would fill their warehouse and they were worried they'd be stuck with the backlog.
Eh, not really. All math / model based predictions are probabilistic. There's other ways to make predictions, and not all of them are based on math, and they might be wrong more often than a probabilistic model, but they exist.
Sure, fair enough, but there are times where a computer model is missing obvious context, and it's those times that I think we have to pay attention to.
The current industry adoption patterns seem to veering pretty close to "the computer did that auto-layoff thing" from Idiocracy in my opinion.
Absolutely not. We need to learn the difference between intelligence and expertise. Is Obama an intelligent person? Of course. Is he allowed to have and voice an opinion? Sure, it's a free country. Does that mean that his opinion is informed by expertise and should dictate peoples actions and therefore the direction of an industry? No.
This is the same logic that allows right wing ideologues to become legitimate sources of information. A causal interest in a topic is NOT the same as being an industry expert, and the opinions of industry experts should be weighted far heavier in our minds than people who "sound like they know what they're talking about".
And your logic is the same followed by government agencies when they effectively agree to regulatory capture because all of the industry experts work at this company, so why not just let the company write the rulebook? 🤔
I personally don't believe we need "industry experts" in every new, emerging type of tech to be the sole voices considered about them because that's how we largely arrived at the great enshitterment we're already experiencing.
Edit: It's really quite a baffling take (given a moment's thought) that the big problem and/or a large problem facing America is that we aren't cozy enough with "industry experts". Industry practically write the policy in this country, and the only places where we have any kind of great debate (e.g. net neutrality, encryption) is where there are conflicting industry concerns.