AI Adoption Rate Trending Down for Large Companies

[Bernardo de La Paz]

I have been expecting a bit of a pause or retrenchment. It is not a reversal. AI is not going away. Even if use goes way down for a few years it will later go way up. The big question for me is whether the wealth generated will go to help reduce the historically unsustainable levels of wealth and income inequality.

AI Adoption Rate Trending Down for Large Companies
Torsten Sløk
Apollo Chief Economist

The US Census Bureau conducts a biweekly survey of 1.2 million firms, and one question is whether a business has used AI tools such as machine learning, natural language processing, virtual agents or voice recognition to help produce goods or services in the past two weeks. Recent data by firm size shows that AI adoption has been declining among companies with more than 250 employees, see chart below.

https://www.apolloacademy.com/ai-adoption-rate-trending-down-for-large-companies/

Chart at the link. Small downtrend starting a few months ago.

[highplainsdem]

1. Don't assume it will "later go way up." As for income inequality - every jndication so far is that AI will

worsen it. Already has, starting with the theft of the world's intellectual property for the profit of a few companies and billionaire tech bros.

We need the AI winter to end AI.

[Bernardo de La Paz]

2. AI is unstoppable. It's foolish to try.

AI won't be the miracle boosters think and it won't be the dud that detractors think. Trying to stop it is like trying to stop email in favour of mailed letters because email "doesn't have the personal touch handwriting does".

[Bernardo de La Paz]

3. Torsten Slok will be on CNBC coming up this hour. . . nt

[Metaphorical]

4. As someone who covers AI in the tech journalism space

Generative (or Transformer) AI - LLMs, media generators, codegen, etc. - will not disappear, but nor will it become all pervasive. It will find a key place in many areas, but it is manifestly unsuited for others that other tech can handle better and more efficiently. One of the big problems the transformer based AI faces is that the cost for individual chats, even a fairly long conversation, is relatively small, but once you start putting it pervasively into various services, the costs add up very, very quickly.

It's this latter point, more than anything, that will keep the current AI architecture from really taking hold. There are a lot of AI startups that failed to account for these costs, and as the costs become known (and as even very deep-pocketed investors begin to fail to see significant returns on their investments) this will be as much of a scaling issue as hallucinations and growing legal resistance.

[Bernardo de La Paz]

5. Good analysis and good points. . . . nt