Disregarding the fact that the article is specifically about customer service I don’t see how it has anything to do with market expectations being corrected, the market is still held together by speculative trading and investors hoping to finally convince enough people to pay the actors out of the massive debt they keep accumulating.
Now I do agree that AI works pretty well as a helpful tool when it comes to coding specifically, but only in the hands of a developer who knows what they are doing. I use Claude Code sometimes in my own work as it does a pretty decent job at speeding up troubleshooting but compared to just going to Stack Overflow it’s a neglient improvement that I can live without.
I have Claude running locally on my PC but I only ever touch it once I have exhausted as many resources from the free plan online as possible
I am referring to the industry possibly using AI - AI makers is a different topic.
Duolingo is a good example how a company wanted to cut costs and consumers said no to it.
Also AI was hyped by AI makers as this amazing tool but reality showed besides to the cost explosion a different reality.
Data center issue aside - in robotics AI will play a massive future. Looking at robotics and Asia as it shows how far at times the west seems to have fallen behind.
This is what I mean with AI isn’t AI - it is a large field.
What will be interesting to see is if and when how the AI bubble will burst and what the outcome will be.
What openAI and Anthropic burn in investment money seems crazy - but hey investors also throw the money at them.
Chinas counter move of open weight models is interesting but I cannot escape the feeling that the CCP in the background funds this a lot to keep it afloat.
I did use Qwen 3.7 local a lot and then 3.8 max online because it is also free and thus allowed me to test it as desired.
Inkling is a fully open source approach but not mean to be Claude level. But the increasing cost of claude and specifically Fabel put a ban onto their in some companies.
Thats why I was surprised to see how the underdog Meta came out of nowhere with their 1.3 AI model.
For coding it is stated they they got so good because many developers selected the contribution model which gave basically free access and allowed meta to train on your prompts.
This will be interesting to see if other companies will follow.
A recent video I watched stated that ai is trained on only 4% of human knowledge - the rest is behind paywalls.
Asking google Gemini about sources often now can give your Facebook Reddit and such which is quite amusing to say it politely.
At school when a students does not cite sources we have a good indicator of probable AI use case.
The idea of AI is destroying the internet is somewhat valid in this context.
Seeing how design students are concerned about AI is nicely refreshing.
.. I quite like that phrase, i find it tragicomic, even though it also makes me wince:
“Science never promised a better world; only truths.”
Very “funny,” isn’t it…?
I find myself thinking : what if we applied that kind of thinking to “AI”, to this screen that’s burning my eyes, for instance, because I’ve taken an interest in 3D modeling via Blender, instead of learning how fruits and vegetables grow (we leave that to others, placing all our “trust” in them). In short, if all this “hardware” were reduced to its proximity to “Science”, AI included… “Truths” rather than a “better world”: is that possible?
What would it look like?
Well, apparently we don’t have much to fear on this front yet, given the quantity of lies still in circulation.
The Chinese short drama video I posted has a few interesting aspects to consider:
It is mass produced soap opera
It seems there is a market form it
AI did take a big hit on it because now actors can be replicated digitally
Seems consumers are fine - well it is soap opera stuff
Would a Nolan use that? No!
He produces a different material
Life of Pi was an amazing movie they won an Oscar and everybody was fired - for what ever real reason.
Was it only production cost?
Going back to the Chinese mass produced stuff an area where AI can be of use is doing tasks that normally in such a production would be too expensive.
Mandelorian and the led screen to cast correct reflections vs Black Widow with terrible actor and scene lighting differences are a good example to show how for such tasks AI possible can do a better job - if AI gives you the control to do it! But it will still be a human to do it and making the decissions.
My point is that with all the down sides AI as a tool can also enable you to do work normally impossible (time cost etc).
The work I do for the Henry Ford Museum started to include AI too. It cut cost for them while allowing me to deliver an even better product in shorter time.
Actually more profit for me.
And maybe at one point they can do it in house alone. It will hurt me but allows them to do work impossible before.
A museum isn’t swimming in cash.
The work I do at the DIA is the opposite AI has zero ability to do anything there - all hand craft.
And the training I do at companies form Czarnowski a titan in trade shows to others show how they use AI - mainly ViacomAI for very specific tasks allowing them to produce work faster and better than before.
Frame to frame or image to image consistency is still an issue. When they is fine 3D rendering will be different.
Also, a lot of people died of hunger and back braking factory working conditions.
Also, I tested it myself and it’s shit. Simple thing really: a scifi crate. The result is OKish when rendered from afar, but when you look at the thing it made it has so many issues that it’s not worth the trouble. Also, it didn’t really control blender and modeled the thing. It wrote a script to assemble the model from basic shapes and slapped on some bad looking material on everything.
I provided some examples - you can also use Google …
it logically is not like AI reduced 10 jobs here but produces 10 jobs there
how this all will play out we will see only afterwards - not before
I have in mind another kind of “tools” that helps on cutting costs:
pay your workers just on net part of the salary…at least in EU (or better, in Italy) it’s a 30% cuts on costs…not bad
if you do a physical job (I mean not digital), you could steal materials or buy it on the black market..you can avoid a 22% of VAT (at least in Italy) and, as the supplier it saves on taxes, you can ask for a further discount…at least another 30% of reduction
I think that they’re too conservative and too little progressist, they’re missing a world of opportunities.
Far to be defender of AI, something that could imagined about new job positions are related to the building phase of new data center…so workers for building company, electricians and plumbers for technical equiments, few IT engineers… but how much? and how long will be these new positions? ok, there is maintenance..but it’s just a minor part of the building phase…
Nothing about AI makes any part of 3D modelling quicker. AI makes horrible topology, lumpy meshes with way too much density, sloppy UV maps. We briefly tried using AI models at work- the amount of time necessary for cleanup dwarves the time it takes to just do it right in the first place
I was thinking specifically about the times that I personally have asked this question, and received no answer.
It’s a regular occurrence on Twitter.
Ironically, even now you’ve told me to go use Google – which to be honest, the number of times I use a Google search in a month can probably be counted on one hand.
I have no reason to believe that whatever I would find on Google, would be a list of wishful thinking likely written by an AI advocate without much real reality basis to it.
The list you provided above regarding the robotics, new chips, AI involvement in image - a lot of that is not a new job, and all of it are tasks that the AI Chatbot companies would love to see automated by AI.
Any jobs related to the construction of a data center are going to be temporary at best, and the number of people involved in running the data center in the low two digit number.
