AI putting everyone out of a job

[Gibbon]

AI is erm artificial and although very good at regurgitating stuff already known is never going to replace genuine creativity. Certainly the days of clerical and call centre staff and even advice givers such as GP services are numbered but extending the knowledge pool will still be in the hands of capable people. However I suppose give enough monkeys enough time at the typewriter and they'll produce something of interest so it is possible that with enough processing AI will create something new but will anybody recognise it as such? The safest careers are the ones requiring physical labour and creativity.

To look for a 'safe career' has always been a slightly silly think to do. Things are always changing. I'm finding AI is just a faster stack overflow at the moment, allowing a forgetful old fart like me to see the syntax I need to help solve the problem. I have very loose requirements so I have quite a lot of ownership, which will be refined with the customer once prototyped.

[willendure]

Don't overlook the fact that using silicon based digital chips, creating a computer with roughly the same computing power as a human brain will need 10MW of electricity to run it compared with the brain itself which needs only around 10W. The computer will run 24/7 though, and never get tired. But with todays digital silicon substrate there is a power factor of 10^6 before AI outcompetes us.

I do think it will happen and possibly sooner than we might think: https://www.extropic.ai/

[willendure]

https://www.youtube.com/watch?v=CNw443X6dB0

"Takeaways:
1. Dual-Speed Tension: Execution is getting radically cheaper, yet the resulting security and quality nightmares create an equally large wave of new work.
2. Infrastructure Gold Rush: Sky-high GPU and cloud bills are spawning lucrative roles in tuning, arbitrage, and large-scale model deployment.
3. Trust Is the New Moat: PMs, UX designers, and leaders who can build user and stakeholder trust amid AI-driven chaos become indispensable.
4. Accountability Outlasts Automation: Program managers, QA, and compliance pros thrive because LLMs can’t own outcomes—humans still must.
5. Data & Retrieval Talent Crunch: Data scientists, vector-database engineers, and RAG specialists are in shortest supply as enterprises scramble to prep data for A
I.
6. Security Arms Race: Red-teamers, AI psychologists, and jailbreak hunters will never be idle while prompt injection and model exploits keep evolving.
7. Three-Step Career Playbook: First automate your own drudgery (survive), then layer on complementary AI skills (adapt), and finally create frameworks or standards
others adopt (lead).

Quotes:
“We’re stuck between execution getting faster and the quality nightmares that speed creates.”
“People get paid to solve problems—AI just moves the problems to new places.”
“The heart of any role that survives in AI is accountability; LLMs can’t take that.”

Summary:
I outlined four dynamics reshaping work: execution is accelerating, speed spawns security and quality nightmares, compute costs are exploding, and the human-AI bound
ary is unresolved. These forces are rewriting 15 core tech roles. Product and program managers must blend AI fluency with trust and accountability. Engineers will cl
ean up vibe-coded messes while DevOps, data, and infra specialists tame runaway GPU bills. UX, security, and customer-facing teams must police the human-AI interface
. Future roles—agent-fleet orchestrators, context suppliers, AI compliance experts—loom. The path forward: automate to survive, add technical depth to adapt, and bui
ld new frameworks to lead with clarity and confidence.
"

[sadkingbilly]

you're all DOOMED!

[Snooky]

In my opinion AI will be most severely limited by energy constraints, unless either the computing power needed to train and run LLMs etc significantly reduces (through more efficient hardware/software), or cheaper, cleaner forms of energy increase.

For example, the IEA estimated that in 2022 data centres, AI and cryptocurrencies consumed around 460TWh of electricity, almost 2% of global electricity demand. AI advances (e.g. video generation) and much more widespread usage in the few years since then mean the figure today is probably significantly higher and only likely to go up.

[willendure]

In my opinion AI will be most severely limited by energy constraints, unless either the computing power needed to train and run LLMs etc significantly reduces (through more efficient hardware/software), or cheaper, cleaner forms of energy increase.

For example, the IEA estimated that in 2022 data centres, AI and cryptocurrencies consumed around 460TWh of electricity, almost 2% of global electricity demand. AI advances (e.g. video generation) and much more widespread usage in the few years since then mean the figure today is probably significantly higher and only likely to go up.

It will come down massively - did you check out https://www.extropic.ai/ ?

On the other hand, even if it does come down massively that might just mean we use a lot more of it and the overall energy usage goes up. But it will mean we can have AGI super intelligence in our phones.

[NotAllThere]

My take on this is Gödel's Halting Problem.

For AI to produce good code productively, it has to be well specified. This is called programming...

There are many programming tasks that don't need creativity. The code monkeys at this level will be replaced. But where real creativity is required, AI is unlikely to replace us before I retire.

I use AI (specifically copilot) to replace search engines and to remind me what the library I need is called.