Goldman Sachs has a big note out on AI’s effects on the job market so, obviously, we asked GPT-4 to rewrite the executive summary as a sonnet:

The rise of AI has brought to light
A question that’s worth a fight
Will it automate tasks with ease?
And drive labour costs down to please?

Generative AI’s potential is vast
It can create content that’s unsurpassed
It can break down barriers between man and machine
And bring about a new era that’s never been seen.

But if it delivers on its promise
The labour market could face a crisis
Two-thirds of jobs are exposed to some degree
And generative AI could substitute up to one-fourth of current work, you see.

But history shows that worker displacement from automation
Has been offset by creation of new occupations
The combination of significant labour cost savings and new job creation
Raises the possibility of a productivity boom that’s worth celebration.

Generative AI could raise annual US labour productivity growth
By just under 1½pp over a 10-year period following widespread adoption, we both knowth
Although the boost to labour productivity growth could be much smaller or larger
Depending on the difficulty level of tasks AI will be able to perform and how many jobs are ultimately automated, that’s the wager.

The world is changing at a rapid pace
And we must keep up with the race
Technology is advancing every day
And we must find a way to stay.

The future is uncertain and unknown
But we must not let our fears be shown
We must embrace the change that’s coming
And keep our minds from ever numbing.

The world is full of endless possibilities
And we must take advantage of these opportunities
We must work together to make a better tomorrow
And never let our dreams be drowned in sorrow.

So let us take this journey together
And make the most of every endeavour
For the future is ours to create
And it’s up to us to determine our fate.

. . . which is, let’s face it, rubbish. The meter’s all wrong, what little information it includes is buried in contrived doggerel, and “knowth” isn’t even a word. AI having failed to make broker research readable, the rest of this post was written by a human.

Goldman’s sift through jobs databases suggests about two-thirds of occupations are exposed to some degree to AI. Of those, a bot can do between a quarter and a half of their workload:

Anyone losing 50 per cent or more of their daily workload to a bot will probably not have a job, Goldman assumes. Its focus is on the desk-bound, as the baseline scenario doesn’t include AI-enhanced robotics:

Disruption is therefore more of a DM theme than an EM one:

AI’s productivity boost may be twofold: employees will spend their time doing more valuable work, and those that are displaced will eventually be re-employed. As a reason for optimism, Goldman cites a 2022 study by economist David Autor et al that found “60 per cent of workers today are employed in occupations that did not exist in 1940, implying that over 85 per cent of employment growth over the last 80 years is explained by the technology-driven creation of new positions.”

Goldman’s base case is for 7 per cent of workers to lose their jobs entirely in the decade after generative AI reaches half of employers, but that most will find something nearly as productive to do. That’s set against broad improvements in workforce efficiency that can add around 1.5 percentage points to US labour productivity — which would be about double the current rate.

The AI growth bonus globally may be 1.4 percentage points, representing almost $7tn in extra annual global GDP over 10 years, it estimates. There are lots of variables involved in those estimates, however:

Goldman clients can read the full note here; the rest of us can refer to this GPT-4 summary of it as haiku:

AI’s job impact, Productivity growth raised, A workforce displaced.