There's a lot of excitement around generative AI, with many people and economists getting caught up in the hype. However, it's essential to take a step back and consider a couple of things. Firstly, the Gartner hype cycle monitor shows that the technology is currently at the "peak of inflated expectations," which suggests a potential downturn in expectations. Secondly, Hofstadter's law reminds us that estimating the time it takes to complete difficult tasks is always tricky. Just because there's a lot of buzz around something doesn't mean it will have an immediate, sweeping impact.
The Economist recently published an article titled "A short history of tractors in English," which drew parallels between the slow adoption of tractors in agriculture and the potential slow uptake of generative AI. The lesson from history is that even groundbreaking technologies can take time to make a real impact due to various factors like usability, labor market changes, and the need for adaptation.
However, there's one area where generative AI could have a more immediate effect: computer programming. Traditionally, programming has been a complex skill requiring mastery of various programming languages. With the emergence of tools like ChatGPT, there's a new possibility – non-programmers can use plain English prompts to instruct machines to write code for them. This could democratize programming and make it more accessible to a broader audience.
Despite concerns about the future of programming in light of these developments, evidence suggests that programmers are embracing AI assistance. A survey of software developers found that a majority are using or planning to use AI tools, seeing them as a way to boost productivity and improve coding accuracy.
In conclusion, while the transformative potential of generative AI may not unfold as quickly as some expect, it could have a profound impact on the way software is developed. This shift may require software engineers to adopt a more engineering-focused approach, moving away from the artisanal craft that programming has been traditionally seen as.
What I’ve been reading:
- Gary Marcus's Substack blog discusses AI companies lobbying for exemptions from copyright infringement responsibilities.
- Diana EnrÃquez's piece on the Tech Policy Press website explores what it's like to be "managed" by an algorithm.
- Margaret Atwood's Substack post delves into films about the French Revolution, starting with Ridley Scott’s Napoleon.