You still get the same issue, I don’t think Alice would lose her Job in this scenario. Who is going to deal with long winded financial tax queries? Is an AI able to sit in a meeting with all the other department heads and answer long winded questions relating to that business?
Are you suggesting that a business can run without someone well versed in Accounting? Just like my example with the shoe business someone will still be needed to oversee the accounting part of the business and serve as a point of contact for all departments.
I get the feeling you don’t really understand where A.I. is going. Give it some years and there won’t be a single human on earth as “well versed in Accounting” as the A.I. doing that.
I will give you a simple example:
If you today want a music score for the next hollywood blockbuster, a world class componist will take weeks or months (Hans Zimmer, Ennio Morricone, John Williams) etc. pp.. And they will proably cost quite some money to do it. Then you will hire an orchestra and rent an opera house or large scale recording studio to play that stuff, singers, mixers, sound technicians to record that. You will need (and pay) dozens if not hundreds of people until the finished track is ready to be used in your movie. It will cost you millions over millions for the soundtrack alone.
In some years you will need a team of maybe 5-10 A.I. experts that will specify and oversee a project to let an A.I. compose such a movie soundtrack and it will have finished the job (including recording, creating OST-albums, merch and cover artwork) within hours. Then you will need a few days for the “polishing”, overseen by your small crew of human listeners, checking if it fits what the movie maker wanted. And you will not hear a difference between that outcome and what a top-class human composer would have given you.
I’d imagine it would take time as a human composer would have to see the film in order to create music for it. I don’t think AI could do it any faster because I believe the music and sound are the last parts to be completed on a movie so time might not be an issue.
Hmmm, you seem to be ignoring a few things, take Hans Zimmer for example who did the music for Interstellar, If an AI could be programmed to make music that perfectly fits a blockbuster it would most likely cost more than a producer/orchestra, also the 5/10 AI experts you mentioned, wouldnt they just be people like Hans Zimmer but with musical AI programming skills? Or do you think any AI programmer can come along and create a program to create music for a blockbuster? The most realistic scenario is that people who make music for films will just have someone on the team who is well versed in AI music creation.
Also you’re forgetting, If the big composers like Hans Zimmer has a proven track record of creating music that fits a blockbuster, why would a producer ignore all human composers and their 1000s of musical scores and soundtracks in existing movies?
Again I come back to the same type of question, If a movie director has no knowledge of how to create music for a blockbuster but has ideas on what he wants I think he would still hire someone like Hans Zimmer so that there is someone there who has a deep understanding of music and then it would be up to the Hans Zimmer person to use AI or not. I think it will still work this way in the future.
Ultimately it’s difficult to ignore the existing music producers as they have already proven they can create music that people will pay for, the amount of money the movie grosses is testament to their skills.
Electronic music is big, so i’d imagine there will be apps created using AI that can create music, also I’m sure the music industry would embrace these apps just as they have many other forms of electronic music creation, with some industries AI will definitely become part of the toolset used to achieve the goal, but will it completely take over the industry? that’s doubtful but debatable.
Elderly care is already being automated in Japan, successfully so, simply because there are not enough people who can work in this field.
Construction can be automated by using 3D printing, by using prefabricated building parts instead of laying brick on brick. That’s already being used simply because it saves ungodly amounts of money and time and reduces maintenance cost in the long run due to standardization.
Crafting jewels is a simple matter of cutting stones in certain shapes. Computers are better at recognizing these shapes than humans. Or cast metals into certain shapes. AI can do that too. At some point, there aren’t humans needed for those cheap jewelry ripoffs needed any longer. It’s already starting here. Combine this with the already used machines to produce necklaces, you can mechanize the entire process without more than a handful of humans for millions of customers.
Tourism and hotels already use lots of automation and AI-like features to reduce the number of humans needed for tasks.
Every single job in our economy can be heavily automated, replaced by AI in one way or another, or mechanized or robo-optimized in ways that make most human work unnecessary. That will displace enough jobs that most people will eventually not be able to be employed. Not by fault of their own, but simply because they are not needed anymore.
Luckily my field and my friends field isnt so easily. So my field yea there are more robot massage machines popping up, but I dont think it will ever have the nuance of an actual human providing therapeutic bodywork.
As far as my friend, field service technician, someone still has to fix the machines on site, or move machines around. Like currently he contracts and is replacing all the photo printers at walmart. Gettin in and out of floors, cable management, going to each device and getting certain steps in the process, will alway require a human that an Ai as of right now cannot replace. Unless we have actual androids walking and moving like we do with the technical know how
I have to admit Chat GPT is a good tool which has helped me achieve my goals. I buy and sell computer stuff on Ebay and up until recently I was using the existing graphics tools on Windows 11 to create thumbnails for the products I’m selling.
With a a text statement I can manipulate a picture to my liking which definitely saves me some time, I am definitely considering upgrading from the free subscription to the standard for £7.00 per month which is cancelable anytime. I would think lots of people who do what I do would also upgrade but some people might be good with graphics tools and enjoy doing it themselves.
Perhaps they could look at a cheaper monthly price where it would be easier to lock me into a contract, many people might only purchase the service for a few months at a time, they will pay the first month then immediately cancel so if a need for Chat GPT doesn’t arise then it won’t be renewed.
I guess they will need lots of people subscribing for a few months at a time to make profit, as of July 2025 there are 35 million paying chat GPT users, so if all of these users are on the cheapest plan then ChatGPT’s income will be around £3 Billion per year.
It looks like ChatGPT’s outgoings are around $700,000 per day which is around $250 million a year so it looks like they are doing well.
I just found an Asus x99 E WS motherboard in one of my pc cases which appears to be worth £150 on Ebay. It would seem AI has increased the value of these server motherboards which can power multiple GPU’s and features Quad memory channels.
There doesn’t seem to be many available in the UK and the ones on Ebay from China seem to have a large watchlist and active bidding.
My original plan was to purchase this machine and add some 5090’s to it so that it’s paralell processing power could be rented online to people training AI models, upon further research I found it wasn’t very profitable as electic is high in the UK. The 32GB (4x8GB) of DDR4 server memory seems to be rare as each module seems to be worth around £169. I didn’t pay very much for the PC so it looks like I will have almost tripled my money just by investing in hardware once all of it is sold.
A good thing about AI is that it’s found a use for old server motherboards and older Nvidia GPU’s.
No wonder, currently AI is on the level of a toddler, barely beginning to develop skills. And of course the companies selling them are exaggerating their competence.
Do that study again in 20 or 30 years, which is less than one generation of worklife. As said in the video: AI is already pretty good in creating images, audio, video. Thats only the first steps.
Yeah right now you can’t fire 100 humans to replace them with 3 Supervisors and an AI license. But thats not the point. The tendency is there and the success rates will go up over the years.
I think what you’re talking about is the concept of using parallel computing in a consumer product. Parallel processing has been around since 1957 primarily used by scientists, the only reason for it being developed was to overcome the bottleneck of sequencial computing. AI within consumer products will only be as good as the people designing them.
My point was, what is it that you expect will change in the next 20/30 years that will lower the failure rate of current AI consumer products?
When I say consumer product, it also includes an AI system designed to replace a person in their job.
Do you think we will learn how to program better applications with AI/Parallel Processing for example?
Say we have a business like Amazon where currently they need people to handle stock which will include storing and retriving and dispatching stock when it sells. Based on the coldfusion study if AI replaced humans in Amazon they would lose a lot of money.
In 30 years anyone NOT using AI for that will be completely out of business. Yes, a few humans to oversee the process and step in when the AI can’t solve a specific situation will still be needed. But 99% of the current workforce won’t be needed any more. And that is where the jobs will go away, you simply won’t have the need for humans any more if robots controlled by AI can do that task tirelessly, without strikes, demand for payment, healthcare and human rights.
The current testing results are not the slightest representative for what AI can do. It’s like claiming in 1950 that nuclear power is totally unlucrative.
I think everyone understands that AI is a very powerful tool when applied correctly. I think the current testing results serve as an indicator which shows how effective AI is when applied to a real life job situation in it’s current forms.
Going back to the Amazon example, yes it quite likely a robot can be made that can replace some small human jobs, but even in 50 years time I can’t see Amazon having a 100% robotic workforce the amount of research & finance and labour needed for such a robot would be too expensive.
In the Amazon example, a human worker would notice small problems in the product they are about to dispatch and swap it out, to program a robot or AI to do this for millions of products simply wouldn’t work and would cost a lot to research, ultimately this kind of project would be abandoned due to cost and understanding that it will always be better to have a human looking out for product defects. Even in the future the cost of AI mistakes when shipping amazon orders will just be to high, the AI robot would cost a lot and then rectifying mistakes would cost even more.
I think 99% of the people have not the slightest clue how much AI will change our socities and business models. AI will change the world faster and harsher than the automobile or the airplane did.
Again, nobody talks about 100%. We are talking about a crew of 50 supervising a local delivery center that works 95% automated. Where you now have 1000 employees to do that you will then be able to deliver more stuff to more people with 950 less human workers.
I think most volume suppliers already use automation to store and retrieve products. I worked for a computer supplier around 25 years ago and they were able to use an automated machine to store and retrieve computers, the machine could handle storing a good 300 computers. There was never any issue, a technician would take the build as far as they could and if they were waiting for parts delivery for the build they would put the half completed computer on a conveyor and instruct the machine to store it in one of the 300 bays. The machine did this 100% of the time without issue.
You’ll find many computer related suppliers already have a high level of automation and they still seem to need lots of human staff. If you’re saying the likes of AMD or Intel could fully automate (95% or whatever % you think) in the future you would be wrong. I’m not sure humans can be replaced easily, if it were to happen the cost would far outweigh any benefit and it seems that there are many people in the world who are happy to finance this fantasy.
If you need billions of numbers processed then yes AI is the go to, if you need products managed and shipped then you need a human AND a sophisticated level of automation if you’re dealing with millions of orders.
You keep focusing on “physical business”, but thats like claiming the invention of airplanes hasn’t changed medicine that much. It’s not the purpose nor the claim that AI (especially todays AI) can totally replace all physical work. At some point AI controlled robots will do most of the physical work, but thats beyond our lifetimes.
What you will see is parts of the business world - and I talk about things like media creation, translation, advertisement, entertainment, design (aesthetics, not functional in the beginning) will be done by AIs for the most part and an overwhelming number of people currently working there will not be needed any more. AI will be able to compose music, not worse than a human can do that. Many movie creators won’t hire human musicians any more if an AI can compose the score for 1/100th the price, and faster. With barely a noticable difference. Yes, AI today can’t do that - yet. But it’s already good. Not as good as a top notch human, but thats only a matter of time.
And that goes on to poster and logo creation, translation models, storywriting, game design. And so on. All flawed today, but the better trained the AI is, the better the results and it doesn’t take that much to become better than a human in most of those things.
And it’s a total mistake to believe that in order to oversee an AI you need a human that is also an expert on that field. Not in the art or creativity department. A producer can absolutely listen to 20 AI generated songs (which the AI can spit out in an hour) and pick the best one to go on with. He doesn’t need to be a musician himself. Not even counted that a human musician would have needed a year to present 20 songs to pick one from.
AIs will be able to check 1000s of Xrays for signs of cancer each day and you would today need hundreds of Doctors to do that. And then only a handful to do a plausability check for cases that are unclear - which is MUCH faster and will return much better results than what a 100 human doctors will do today. This is the power of AI, not controlling Amazon stores.
Yes, all of what you say is 100% correct. but will the produced film sell well? I’m not so sure. People have studied hard for many years to be able to create/compose music which is a big part of the reason why we like the musical scores in many films. Do have a look at how music for a film is created and you will find that it will never be as easy as a complete novice using AI to make a sound track.
The AI is just going to copy the music that is already out there and then make it’s own variation of existing music.
If you have a look at how a professional creates music you will see that he is using lots of tech created specifically for that purpose, the producer tries to match the music to the emotion happening on screen. This is a skilled producer who understands exactly what people watching that movie or show want to see. If you’re telling me that an AI will one day understand emotions better than a human can, again, I would say you are wrong. The understanding of human emotion is crucial in making a successful movie or show.
Yes an AI can definitely copy whats already out there and build an understanding of what’s required I can agree, the only problem is that the operator would be a novice with music and can barely recognise musical notes like A B C, if the AI makes a mistake or times the music on 20 of the scenes wrong again, are you saying a complete novice can recognise all of this and change it? would the producer have the time to micro manage what the AI has created?
In this particular field I’m very sure producers are happy to pay a music producer to oversee the music, A music producer would be able to control the AI music tools many times better than a novice.
I think there is an AI created scene in an upcoming movie where Will Smith is fighting a ball of spaghetti, what’s to be understood is that you still need a fight choreographer and people with a good understanding of a fight scene.
