How 19th century luddites differ from today's political luddism that tries to bound AI?

Question

There are ongoing preparations for the AI regulation summits https://www.politico.eu/article/eu-us-uk-china-artificial-intelligence-control/ that are planned to be held by G7 AI control group and another summit that is being organized by UK - both are planned at the end of 2023. And both more or less explicitly includes "job security" as one of their concerns. Essentially "job security" means that AI is being bounded to preserve the existing need for human workers, by not allowing AI to achieve the necessary capacities or by not allowing AI to use its capabilities.

My question is - how such bounding of AI differs from the 19th century movement of luddites. I am asking for the essential, theoretical differences.

Of course, there are differences from the point of view of class struggle. 19th century luddites were against the interest of the ruling class and the 21st century AI luddites are trying to preserve the upper middle class jobs, class that is politically most active and influential. So - luddites have different power structure now with the different outcome of their fight.

But if we inquire of the essential feature of both movements, then we should ask what is different now. And if current technologies are really that different, then maybe we should not bound them, but - instead - we should try to solve the problems whose solutions were deemed unsolvable with the old technologies - e.g. scarcity.

I am reading, of course, the best book about AI https://www.nber.org/books-and-chapters/economics-artificial-intelligence-agenda But I would like to understand why one movement of luddites is perceived to be so bad and another movement of luddites is perceived almost as inevitable? And is there any justification from the viewpoint of economics for this? I can imagine, that there can be political or emotional justifications, but those are for other discussions.

Answer 1

As you said both classes of luddites, those of the 1800s and those of today, had or have a job security concern. I would separate "job security" concerns into three issues common to both eras...

1. job displacement involving a worker making a change in job or career

2. disruption to existing job processes that require adaptation or training or education

3. fear that there won't be enough jobs

There is the possibility that workers can successfully navigate a displacement or adapt to a disruption to existing job processes. If displacement and disruption enable a general increase in welfare, they should be welcome. Data not specific to the introduction of AI shows that as technology was adopted over decades, median real household income increased, in the context of displacement and disruption.

Data not specific to the introduction of AI, shows that as different forms of technology are adopted over decades, prime age (25 to 54 years) labour force participation has not suffered so the worry that there will not be enough jobs is not supported by the past 70 years of data. A person might say that an increasing trend ended in 1987 and since then it has fluctuated around 82%.

It may be that these workers are also enjoying more leisure. This is uncertain because average weekly hours (worked) is not reported as a prime age measure. This graph of average weekly hours (worked) supports the idea that there are more leisure (non-work) hours per week...

You might say young adults in school are at leisure because American universities are increasingly like country clubs and retired people are at leisure so look at this graph that suggests young adults spend an increasing number of years in school and data showing old people spend an increasing number of years in retirement...

(Edit... It was suggested in the comments that the leisure data is only weakly related to my claims. Please see my post in the comment section about how technologically enhanced work should be welcome. )

Additional to that there seems to be speculation. Some examples follow.

1800s luddites differ from AI luddites in that AI luddites also have a physical security concern. For example people fear a paperclip maximizer.

AI luddites also have a political regime concern because AI might manipulate voters. It is believed Prigozhin used human workers to manipulate voters but conceivably AI will be able to do it. A reference on the matter of Prigozhin manipulating voters.

AI luddites also have country ("national") security concern because the route to paperclip maximization might involve war or AI manipulated voters might empower leadership that leads countries to war.