Lightnews — Scholar-powered news

Chris Kanich

@kaytwo.org

130 followers 130 following 50 posts

I get unreasonably excited about Chicago, cybersecurity, education, and local impact.

https://kaytwo.org

Posts Replies Media Videos

Chris Kanich

@kaytwo.org

A lot to chew through in this fantastic piece by @randomwalker.bsky.social and @sayash.bsky.social knightcolumbia.org/content/ai-a... The models exist now as "Normal Technology," how we use them and how society adapts to them is where the impactful and interesting work is happening.

Image illustrating the stages of AI impact, combining a flowchart and a detailed table, with a caption and citation.

Top Flowchart: Shows three stages in boxes. "Invention" (blue box) leads via an arrow to "Innovation" (purple box), which leads via an arrow to "Diffusion" (orange box). A curved arrow loops from "Diffusion" back to "Innovation", indicating a cycle.

Table Below: Organizes information about AI impact across four stages corresponding roughly to the flowchart.
Rows are labeled: "Stage", "Example", "Appropriate metrics", "Speed limits".
Columns represent the stages:
1. **Methods / capabilities** (Blue background, relates to Invention): Example: LLMs; Metric: Benchmarks; Limits: Herding, "Last mile" problem.
2. **Products / Applications** (Purple background, relates to Innovation): Example: AI code editor; Metric: Uplift studies; Limits: Capability-reliability gap, Dependence on diffusion.
3. **Early adoption** (Orange background, part of Diffusion): Example: Individuals creating simple apps using AI; Metric: Surveys; Limits: Learning curves, Safety.
4. **Adaptation** (Orange background, part of Diffusion): Example: Retraining software engineers to emphasize AI skills; Metric: Economic indicators; Limits: Organizational changes, Laws and norms.

Figure 1 Caption: "Like other general-purpose technologies, the impact of AI is materialized not when methods and capabilities improve, but when those improvements are translated into applications and are diffused through productive sectors of the economy. There are speed limits at each stage."

Citation (bottom right): "4. Jeffrey Ding, 2024. Technology and the Rise of Great Powers: How Diffusion Shapes Economic Competition. Princeton University Press, Princeton."

April 17, 2025 at 3:39 PM

Chris Kanich

@kaytwo.org

RETVRN

April 12, 2025 at 7:12 PM

Chris Kanich

@kaytwo.org

AI can definitely be contributing, but imo this is more a reflection of org structure. A BLS “programmer” is effectively a subset of a “software developer,” so during layoffs, who would you cut first? Can see that effect in their dot com bust data.