Rob says we must learn everything we can about AI to remain relevant in the enterprise world and I tend to agree.
As a 35 yrs old employed 100% at a job which will soon become irrelevant, the field of AI is intimidating due to the diversity of the offering when it comes to choosing a skill to learn. The barriers of entry seem high and time consuming.
Coming from a non tech background its hard to differentiate quality courses from fluff and money milking curriculums.
What skills should I really focus on at first? I’m not going to start a college degree in computer science.
How should one get started effectively? What educational provider to rely on? How to distinguish quality and relevance from money potholes?
it’s a hoax and a bubble that it’s about to collapse, if the AI can take up your job then your job is not something important,
what most companies do is to use the AI as an excuse to layoff and hire cheap labour from India or something similar or to get some “innovation” start-up grant
those that indeed fall into the AI hype end up using many programmers and “specialists” to tune and filter continuously the garbage that AI yields
instead of moving to mastering AI move to some hard skill, maybe for example fixing broken computer hardware would be better, if you want your own business you could be buying batches of broken ones cheap and sell them fixed
I reckon that there is an imminent implosion(both artificial and natural in tandem) of modern civilisation and soon there will be various shortages and much trading of used machines like in the Great Depression that they were scrapping the ships
If you insist on AI take sometime to understand it and search the trends, play with huggingface models and local AI, I believe that “studying AI” in some institution is as much useful as “studying Linux”, everyone learns Linux by himself and on the field if he gets hired somewhere
the basic concept of AI is that it creates a multidimensional space(coordinate system) where it places randomly the objects(tokens) of interest and during training the related objects come closer together, then given an input it calculates which objects have the higher probability to follow the input
My best advice is just keep playing around with any LLM to get hands on experience e.g. duck.ai which beats reading and hearing advice from others.
That said, I say look at prompt engineering like this video
… but AI and LLM is still an evolving landscape and things are changing so much that it’s just faster to watch a ton of videos on YouTube. I mean a lot. You’ll eventually get a feel for things. Much easier than relying on someone else’s view.
I’ve been digging into this for a while and I don’t have an answer as to where things are going but I feel much better positioned than months ago. Much of it is because I forced myself to heavily use LLM.
Let’s approach it more technically, there are several questions to be settled.
Do you expect to be hired or do something on your own?
Being hired is very fluid and based on momentary trends, doing your own business is hard but you can get into various communities like gaming where big corporations fail in many aspects mostly because they want profit and they insert politics and agendas.
This leads to the questions who uses AI, I reckon it is used by common people either trying to find help for their homework or fighting their loneliness or trying to escape the cost of outsourcing an idea of theirs like making a comic/music/movie. The last one has outraged artists that would charge 100 euros per picture etc.
Even in banking I haven’t noticed anything special, it just blocks repetitive transactions and logging from unusual locations based on the IP.
Some industries claim to use it in medical screening and various other “matching” activities like product recommendations and dating.
In critical cases like the medical one the result should always be verified by a human and I don’t know to what extent a doctor for example would prefer to do it completely on his own instead of cleaning the AI’s garbage.
Technically you can go from the lowest level of mathematics and programming to the highest of giving prompts, those always interconnect somehow and there is much dependence on the specific model and the various tuning and patches that can be applied. It becomes a madness as you delve more into the mathematical aspect especially when you try to understand why it fails or misinterprets things.
Last for now, is the power consumption, it can work quite easily as an agent on your local machines but training is complicated and consumes power and needs hardware, so don’t expect to experiment so much with retraining it often at least for large models and large pools of data. Maybe working with smaller focused agents is easier. The first layer agent will be deciding which agent will take the question and that will be a small specialised one.
Even for the big-tech I don’t know for how much more they can be building infrastructure and using electricity with taxpayer money in an already failed financial system, just for people flirting and having virtual sex with AI entities.
I really appreciate the time you took to answer, many thanks to both of you.
Maybe one day I’ll make my own business, for now I’m trying to stay relevant on the job market (EU territory), hence my question.
These are interesting insights. I guess most of us are on the same ride, making our best guess for where this is going. It for sure will depend on one’s vision of the future. Will humanity destroy itself and go back to the dark ages or to some sort of technological dystopia? Or will we transitioned to a more evolve society with tech that will emancipate our creativity? It’s a fork in the road.
In any scenario hard skills will be valued, no doubt about it. That’s why I’m working on getting an electrician’s license this year, something I never planned on doing before 2020.
I watched the engineering prompt video and found it helpful, thnx! I take home keeping a prompt library for future usage.
I find that learning from an institution or a program can help getting the basics down faster as you benefit from a structure on the subject. However there are so many online courses nowadays, especially for tech stuff, its hard to know what they’re really worth.
First go through all the free videos on YouTube , maybe you can find paid ones for free on piratebay as well as PDFs books. Look also at z-lib, https://en.z-library.sk/ .
I took a lot of AI certifications spring 2025. (rant engaged) I can tell you none of them prepare you for what is being sought. The companies were looking for the next tool to save them money. You needed to be a DevOps person or someone who got lucky to be on a project to implement some sort of cluster or pilot solution. I found few companies knew what they wanted so much like with the security era push for people, they wanted only experienced folks. (rant disengaged) What I advise is you play with as much and as many of the AI tools as possible. Build your generic toolsets and know what works and what doesn’t work. Understand the governance and data lifecycles (data hygiene is a good one to know as well). Figure out your target enterprise and us an AI to ask what their pain points are. Build a training solution or practical solution with public data sets. Build your portfolio and get some certifications. I think that AI platforms are going to be out there and the tools that get good results will get acquired. My two cents.
I find that learning from an institution or a program can help getting the basics down faster as you benefit from a structure on the subject. However there are so many online courses nowadays, especially for tech stuff, its hard to know what they’re really worth.
Yes, that is great for established fields or subjects. e.g. electrician.
But that becomes less worth it for new and evolving fields like AI. It’s developing quickly that what you learn from a teacher becomes obsolete.
The other issue you’ll run into is something similar to investing.
Let’s say you know nothing about investing so you’re like
This is so hard. I’m just going to outsource this to a financial adviser.
so you then run into the next issue:
Which financial adviser should I trust with my money? How do I know they know their stuff. How can I even verify without knowing anything about investing.
It’ll be simiar to learning AI. You’ve decided not to learn on your own but now you have to find and pay for a teacher/course.
How do you know they’re good, really understand AI, keep up with the new changes and/or even qualified to teach?
If this is the future, I rather make this a long term study session and lead the way myself. I chose correctly. There’s more I need to cover but I feel like I am a great spot.
he tries to stay afloat in a strong turbulent stream instead of letting the stream take him to a delta where the water is calm(“ride the tiger” as Chinese say)
it’s mostly a philosophical choice of conduct, when you are young you try to keep up and you are enthusiastic with things, when you grow old most earthly things and promises lose their meaning and you mostly laugh but also get disgusted with those eating the bait
Excellent, I’m really enjoying the variety of the replies.
The parallel with investing makes sense. Its kind of like learning a language, it is a slow process and one has to learn to think in the new cultural framework.
The Chinese have a lot of wisdom. Hopefully the delta is in the near distance. Meanwhile, unless you have a reliable floaty, its probably a good idea to learn how to swim IMO.
especially swimming is a very strange situation since even a dead can float but the alive ones “have to learn it”, someone who “can’t swim” will float if he is placed on the water surface while sleeping
I think the key thing is where are you putting your energy and focus on. i.e. what you prioritize will vary from person to person.
For me, I deemed AI something that can have short and long term benefits. Personally and professionally. Also, not likely to be a fad… similar to the internet.
But for say, a privacy phone, I wanted it but wasn’t something I wanted to prioritize right now in terms of time and effort. Yes, I could have spent the time to look into installing some Android OS, etc. but I didn’t want to because I have other projects going on that I am focusing on right now. Hence, the Brax3. I rather fork over cash.
I think that’s what it’s about. What do you deem important and how far are you willing to put energy into it?
For privacy I watched Rob’s recent video (would like more of them) and I am thinking he may support:
openClaw + openCode + Ollama (on-prem or cloud hosted) or
AgentZero + openCode + Ollama (on-prem or cloud hosted)
I would prefer on-prem and the mac-mini is being pushed as the go to option, but I would like to know what Rob would recommend for hardware and linux distro etc.
