The great AI jobs disruption is under way – Automation will reshape tech work, and spark new opportunities
https://www.ft.com/content/cb9ea970-e6de-4daf-aa9e-7a48d5e648c3
The great AI jobs disruption is under way – Automation will reshape tech work, and spark new opportunities
https://www.ft.com/content/cb9ea970-e6de-4daf-aa9e-7a48d5e648c3
7 Comments
From the article
Global boardrooms are deep into the experimentation phase of adopting artificial intelligence in their businesses. Many cost and time-saving applications of the technology, from customer service chatbots to nimble data analysis tools, will endure. On other uses, the jury is still out. Coca-Cola’s first AI-generated Christmas “Holidays Are Coming” advert last year sparked an online backlash. This week Swiss bank UBS revealed that it was using deepfakes of its analysts to interact with clients. Despite the uncanny resemblance, investors may trust a live human more for portfolio advice. Some studies place the failure rate of AI projects as high as 80 per cent.
Still, the disruption is starting to show up in job markets. The tech industry is on the frontline. Last week, Microsoft announced 6,000 lay-offs, including in product management and software engineering roles. In April Duolingo, the language training app, said that it would go “AI-first”. That, in part, means only adding headcount if a team cannot automate more of its work. Last year at least 95,000 workers at US-based tech companies were laid off in mass job cuts, according to Crunchbase’s tally of news articles.
Fears of a further imminent and widespread wave of AI-linked redundancies in the tech sector may be overblown, however. Though references to “AI” are now common on S&P 500 companies’ earnings calls, these are not all linked to immediate cost savings projects or investments. They are sometimes provisional plans to implement AI, which may impress investors, but don’t always come to fruition. In other cases AI-induced job cuts can even backfire. Last year Klarna boasted that AI had replaced 700 full-time agents. But recently the CEO of the payment services company said it was launching a recruitment drive to ensure users could always have access to a live representative.
Getting customers used to non-human interactions is not the only reason why AI integration will take time. Legacy IT systems take time to update. There is also a risk of extrapolating too much from recent tech cutbacks. It is routine in times of economic uncertainty for the industry to slow hiring and lean more on technology to make efficiencies.
Yet the long-term trend towards greater AI adoption across the tech sector is unmistakable, and will reshape the industry. Microsoft CEO Satya Nadella recently said as much as 30 per cent of the company’s code is already written by AI. Openings for coders and application engineers are shrinking in the US. The job postings index for US software development roles on the employment site Indeed is at its lowest in five years.
I suspect many of the “new opportunities” will be hiring people to correct the numerous mistakes AI will make.
> The tech industry is on the frontline. Last week, Microsoft announced 6,000 lay-offs, including in product management and software engineering roles. In April Duolingo, the language training app, said that it would go “AI-first”. That, in part, means only adding headcount if a team cannot automate more of its work. Last year at least 95,000 workers at US-based tech companies were laid off in mass job cuts, according to Crunchbase’s tally of news articles.
1. Microsoft stated this was about having fewer management layers.
2. Duolingo backed off this week.
3. There are many reasons for last year’s layoffs (mostly about ZIRPs), but AI ain’t it.
> Yet the long-term trend towards greater AI adoption across the tech sector is unmistakable, and will reshape the industry.
It is in fact mistakeable, as the author has clearly mistaken it.
Another press release by Big Corporate to scare people into accepting poor pay and working conditions.
AI is a dog and pony show. It’s never gonna live up to the hype.
If you want to secure your job (and not hope for UBI to magically land in your lap) I recommend taking a more proactive approach: I recently wrote a [post](https://upwarddynamism.com/ai-society-you/ai-job-security-career-strategy/) introducing a simple framework to carve out your job niche / define your “home turf” as a human and how to smartly “divide labor” with AI. Maybe that’s useful for less fatalist fellows..
I see no evidence that LLMs can replace engineering staff. Perhaps make engineers more efficient, but that’s still an open question.
Recent papers from Anthropic are very suggestive: showing, for example, that LLMs don’t really learn math, but more of a fuzzy finder that searches embedded text space, as one would expect. Logical “thinking” seems somewhat tangential to what LLMs are trained to do.
Now it may be that subsequent model developments give rise to greater intelligence.
At the end of the day, any model capable of replacing engineers would, I think almost by definition, need to be AGI. In which case this is all meaningless, since all labor would become instantly worthless.
Reshape tech work? It’s nowhere close to replacing me, but can absolutely replace management right now.