676
Microsoft insiders worry the company has become just 'IT for OpenAI'
(www.businessinsider.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 22 Jun 2024
676 points (98.1% liked)
Technology
59197 readers
1991 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
All of big tech is really worried about this.
If the AI boom is a dud, I can see many of these companies reducing their output further. If someone comes along and competes in their primary offering, there's a real concern that they'll lose ground in ways that were unthinkable mere years ago. Someone could legitimately challenge Google on search right now, and someone could build a cheap shop that doesn't sell Chinese tat and uses local suppliers to compete with Amazon. Tech really shat the bed during the last economic downturn.
or more like their user experience was already so garbage, adding AI to it doesn't make any noticeable change lol
I don't use a single Meta product on purpose. I'm sure they scrape my data despite my best efforts to not be tracked online.
I still unfortunately order things from Amazon for the convenience, use Windows for gaming and at work, and occasionally use Google search with heavy boolean search, custom search engines, and browser extensions for filtering out the garbage. I also still use Google Maps and I have an Android based tv where I occasionally watch SmartTube.
Hell I even get Netflix included with my T-Mobile subscription. My wife watches that.
And for now, I have an iPhone SE until it dies and I make the switch to a Google phone or something.
Typing this out makes me wonder what I'm waiting for to find alternatives for this FAANG garbage, but I have no idea how Facebook still exists.
It turned out that it's incredible easy to order as guest at other sides
Yes but I don't want to type my billing details every time I need some thing. I don't want to wait 6 weeks. I don't know if other sites are reputable. I don't want to pay shipping. I like being able to wishlist stuff or store stuff in my cart for later and read lots of reviews on products (I'm aware many are fake).
There's also the fact that nearly every website runs on AWS, so even if I boycott Amazon (I'm sure they'll miss my $100 a month in purchases), I'm still providing them money by visiting the sites that are hosted on AWS. Pretty hard to completely avoid them in this day and age.
Amazon for me has been utter garbage in the last 10 years. Fake products, stuff that is supposedly coming next day comes in 3+ days, customer service is some copy/paste canned answers etc
We order a lot of baby stuff on there. They’ve accepted returns on everything that didn’t work or wasn’t what we anticipated. We can walk to a UPS store from our house and drop it off. Anecdotally, they also have the best deals about 50% of the time on PCPartPicker.
It does take longer to fulfill some orders for us. But others show up a day or three early even though we don’t pay for Prime. I used to work for the post office before they switched to their own delivery, and they would drop off their pallets to us in the mornings to be taken out for the last mile by our carriers. It seemed like that was a better experience. It has definitely enshittified somewhat since their golden days.
Monopolies don't care about the user experience, only profit. The AI doesnt understand the former, only the latter. The continued degredation of the user experience is a likely indicator of an increase in revenue as function of successful application of AI.
Do you possibly mean "The AI evangelists" or something similar?
Like, I could totally understand it in the "software will also include the biases of those who wrote it" kind of way (a la Amazon's failed attempt at automating job candidate search). If the only incentive you're given as a programmer is "make it make money", then yeah, your AI is going to bias towards that end.
Just couldn't tell on first reading
I'm not actually asking for good faith answers to these questions. Asking seems the best way to illustrate the concept.
Does the programmer fully control the extents of human meaning as the computation progresses, or is the value in leveraging ignorance of what the software will choose?
Shall we replace our judges with an AI?
Does the software understand the human meaning in what it does?
The problem with the majority of the AI projects I've seen (in rejecting many offers) is that the stakeholders believe they've significantly more influence over the human meaning of the results than exists in the quality and nature of the data they've access to. A scope of data limits a resultant scope of information, which limits a scope of meaning. Stakeholders want to break the rules with "AI voodoo". Then, someone comes along and sells the suckers their snake oil.
its a function of paying their employees less for more work relatively speaking and extracting more profit from consumers through ads and enshitification in general
I fucking bing'd something the other day to get a better search result. What the fuck google.
Try Kagi. Paid search engines are the future in order to extract yourself from the enshittification of "free" search engines.
If your goal is to get away from this AI shit show, Kagi might not be the answer, according to their own blog.
I will search for a very interesting article you should read, before deciding to give kagi any money.
Edit: found it
There's also the whole interaction in how the CEO treated someone who wrote an article and he wouldn't leave her alone after asking him to stop.
From his perspective I get it, you want to have good press and try to clear up any misconceptions. But how he went about it was very unprofessional and far too pushy.
yeah, that seems rough on kagi’s part. i used the free trial, and then just moved onto startpage because startpage doesn’t make me pay.
I read that stuff a few weeks ago. And the responses and discussion on Kagi's Discord. I'll continue to monitor Kagi's behavior, of course, but for now I prefer Kagi. I get far more relevant results with no advertising noise and as much or as little "AI" assistance as I want.
Google is a cesspool and DDG is simply inferior - worthy, but inferior.
Using Kagi is not a bad decision. After reading a lot of positive things about it and beeing quite hyped, I was so ungently reminded, that every good thing comes with its own baggage of bad.
I thought I share it, so that everyone can make their own decision.
I also tried Metager, a german meta search engine. Sad to report: not usable for me, although I like their club (suma-ev) and donated some money towards them.
For now it's DDG I guess. :-(
I'm not sure there could be any sort of legitimate threat to them, but I could definitely see a Netflix situation playing out. That is a popular upstart temporarily seems poised to take over, but then suffers from extreme levels of interference from bigger players who artificially hold the upstart down while they desperately catch up and then ultimately come at least equal while the Netflix equivalent is mostly a shell of what it could've been.
Never underestimate how much buckets and buckets of cash reserves can overcome even incredibly out of touch laziness when it comes to competing with any start ups. Apple in particular could probably afford to let competitors get a decade ahead and still be able to come back based on the ridiculous amount of cash they have to float their business along with.
Yeah competition won't work in a market where some competitors have such massive amounts of wealth. This is a failure of unrestrained capitalism and it's bad for consumers ultimately.
Not really, unfortunately, because of the sheer mass of the internet the infrastructure to just support the index of it requires massive funding. Even other giants like MS with Bing struggled with this. Short of a radical new way to run a search engine without a massive index, I just don't see it happening.
It's kind of curious to me about search because honestly my Internet world has only grown smaller and smaller. Where I used to use Google to find new websites, I feel like most of my searches on Google are now to search a handful of sites I already know. Ironically if Reddit had a better search function, a lot of my Google usage would fall off as I'd just go directly there, as it's still the best place I've found for troubleshooting support and real reviews of lots of products. A competitor to Google wouldn't really need to index the entire web for most people, but rather a relatively small number of website super giants like Amazon, Reddit, Wikipedia, etc.
Kagi literally provides the same quality Google used to.
They pay Microsoft for access to the bing index
They stopped using Bing a year ago.
You're right, my bad. (Source: https://help.kagi.com/kagi/search-details/search-sources.html)
Kagis ok, it's better than Google now, but not the Google of the past and only if what you were looking for is recent and indexed. If not, it just falls back to the same Google results. Good if you want separation and not having Google know what you're searching for, but beyond that...meh.
Whaddya mean, "if"? Emperor wears no clothes...
AI did boom, but people don't realize the peak happened a year ago. Now all we have is latecomers with FOMO. It's gonna be all incremental gains from here on.
I think the true use case for these AI technologies is yet to come. What most people are doing with the "AI" tools available today is just gambling around. But working with personal computers could be changing fundamentally in the coming years.
A simple control algorithm "if temperature > LIMIT turnOffHeater" is AI, albeit an incredibly limited one.
LLMs are not AI. Please don't parrot marketing bullshit.
The former has an intrinsic understanding about a relationship based in reality, the latter has nothing of the likes.
I can see where you're getting at, LLM don't necessarily solve a problem, they just mímic patterns in data.
That is indeed exactly my point. LLMs are just a language-tailored expression of deep-learning, which can be incredibly useful, but should never be confused for any kind of intelligence (i.e. logical conclusions).
I appreciate that you see my point and admit that it makes some sense :)
Example where I think pattern recognition by deep learning can be extremely useful:
But what I am afraid is happening for people who do not see why a very simple algorithm is already AI, but consider LLMs AI, is that they mentally decide to call AI what seems "AGI" / "human-like". They mistake the patterns of LLMs for a conscious being and that is incredibly dangerous in terms of trusting the answers given by LLMs.
Why do I think they subconsciously imply (self-)awareness / conscience? Because to not consider as (very limited) AI a control mechanism like a simple room thermostat, is viewing it as "too simple" to be AI - which means that a person with such a view makes a qualitative distinction between control laws and "AI", where a quantitative distinction between "simple AI" and "advanced AI" would be appropriate.
And such a qualitative distinction that elevates a complex word guessing machine to "intelligence", that can only be made by people who actually believe there's understanding behind those word predictions.
That's my take on this.
And people will still say AI isn't a bubble.
There is a bubble in AI, AI isnt a bubble. In the same way there was a bubble in e-commerce that lead to the dotcom crash. But that didnt mean there was nothing of value there, just that there was too much money chasing hype.
I think it will hinge on one thing: Will AI provide an experience that is maybe worse, but still sufficient to keep the market share, at lower cost than putting in the proper effort? If so, it might still become a tragic "success"-story.
It's very, very costly, both but the hardware and the electricity it takes to run it. There may be a bit of sunk cost fallacy at play for some, especially the execs who are calling for AI Everything, but in the end, in AI doesn't generate enough increase in revenue to offset its operational costs, even those execs will bow out. I think the economics of AI will cause the bubble to burst because end users aren't going to pay money for a service that does a mediocre job at most things but costs more.
That's what I suspect, too, but I'm not entirely sure in my research so far. The question I am still unsure about: Is it as costly in running, or is the real costly part "just" the "training our model" part? I wondered that, because when I was messing around, things like generative text models could run on my potato PC with a bit of python scripting without too much issue, even if not ideally - as long as I had the already trained dataset downloaded.
Can't really answer the expense trade-off until you look at concrete use cases, something general AI is allergic to...
You've got a great point there, actually
No. They are still capable of pressure typical for oligopoly (censoring out mentions of their competition, tactically buying out things which could help that competition and shutting them down, defamation, lobbying for laws directed against their competition).
Unless that happens too fast for them to realize.
Summary: stick to open source if you want usability.
How does that address web search and online shopping?
web search: Searx-ng & co.
online shopping: only thing i see so far in local big shops is AI chatbots to reduce load on telephone support. I never used Amazon, don't care what bullshit they do.
slag?