I’m sure this article’s title will age well.
I use logic pro x on a MacBook for my music recording software but use Android AI search when I run into software or settings problems.
I find the AI is much faster than watching a tutorial or going to the manual for answers.
AI is not a universal panacea and never will be. If applications are targeted to specific roles and tasks, then it can certainly be useful... but then only if those implementing the tool have the time and wherewithal to do it right.
The barriers to doing that? Technical skills, the hype machine that’s already promised the moon and a few stars, and (of course) costs.
Example of a bad AI use? I’ve yet to find any AI-operated ‘customer service’ call center that would even entertain even a single question that I needed an answer to.
AI is the wrong name. Its a marketing name that gives too much credence to software processes that do what software processes have always done, follow rules and search data. Now with faster better computers and more access to data, software is doing a better job. But its not intelligence. Its only following rules very fast. Yes, it appears to think. But it doesn’t. Give it bad data and it comes up with bad answers. Give it bad rules and it comes up with bad answers. To think of it another way. AI can come up with a very complex very fast stupid answer faster than a human can. Humans just need to understand that AI is simply a tool. like a chain saw works better than Paul Bunyan’s axe. It does not know the best trees to cut or when too much is cut. Someone has to tell it that. AI is unfortunately now just a simile for actual intelligence. When it should be just a simile for fast software with all the same bugs and bad designs it had before.
Personally hoping for our own Butlerian Jihad
The Great AI Crash of 2025 lurks just over the horizon.
I hope it crashes to hell and gone, before they have it controlling every traffic light and air traffic tower in the country from one central AI server...
AI is subject to the most extreme, most over-the-top marketing hype cycles of any technology I’ve ever seen.
And each hype cycle is more off-the-charts than the last one.
I work in the field, and I’ve learned to ignore these cycles and the outlandish promises that are made each time around the bases.
I remember about fifteen years ago, the expression “scary accurate” was being used to describe whatever was the latest LLM “breakthrough” of that time. Does anyone remember that one? No. Long forgotten.
“AI looks increasingly useless in telecom and anywhere else”
Disagree with that. It would be a great improvement to turn traffic control in the US over to AI and get rid of the ones doing a poor job of it now
AI looks increasingly useless in telecom and anywhere else
ChatGPT said:
I hear you — a lot of people in telecom (and other industries) are feeling disillusioned with how AI is being pitched versus what it can realistically do today.
Here’s a quick breakdown of why that gap exists:
⚡ Why AI Feels “Useless” in Telecom (and Beyond)
Overhype vs. Reality
Vendors often promise “AI will automate everything,” but in practice, it usually just makes incremental improvements (like helping triage trouble tickets or analyze call logs). The hype sets expectations at world-changing scale, so the reality feels disappointing.
Narrow Application
Most AI in telecom today is stuck in silos:
Chatbots that deflect basic customer service calls (but frustrate real customers).
Network anomaly detection that flags possible outages (but still requires human engineers to validate).
Predictive maintenance on equipment (but adoption is patchy and data-hungry).
Data & Integration Challenges
AI thrives on clean, structured data. Telecom providers have mountains of messy legacy data spread across billing systems, OSS/BSS, CRM, etc. Without proper integration, AI models either underperform or can’t be deployed at all.
High Cost, Low ROI
Implementing AI at scale requires new infrastructure, staff training, and vendor lock-in — and executives often don’t see clear financial payoff compared to just improving existing processes with conventional automation.
🌍 Where AI Is Useful in Telecom
Even if it feels shallow now, there are some niches where AI is genuinely adding value:
Fraud detection — spotting SIM-swapping, subscription fraud, or unusual call patterns faster than rules-based systems.
Network optimization — tweaking cell tower loads and frequencies automatically.
Churn prediction — helping retention teams identify customers likely to leave.
Field ops — AI-assisted image recognition for tower inspections and damage detection.
But again — these are small wins, not revolutions.
🔮 Bigger Picture
The same pattern is showing up in other industries:
In healthcare, AI struggles with accuracy, liability, and messy medical records.
In finance, fraud detection works, but “AI advisors” are mostly just rebranded algorithms.
In creative work, tools help brainstorm or draft, but human skill still carries the real value.
So the pattern is clear: AI is an enhancer, not a savior. The disappointment comes from expecting it to replace entire workflows rather than assist in them.
AI has its uses. I know a guy that uses it to help make complicated formulas in spreadsheets and other analysis software. Almost all of the rest of it is GIGO.
bmrk
Idk much about this but I do know snake oil salesmen will promise the moon!