Umm. I think the researchers focused on the wrong thing. For AI productivity benefits to really work when it comes to software development, It's another LLM who should be review most of the code generated by an LLM, while a human should just act mostly as a judge, moderator, and final arbitrer.
"In terms of raw code being produced, the results are clear and stark. The introduction of AI coding agents at a firm leads to a 30 percent increase in total lines of code generated, a 20 percent rise in the number of total commits, and a 23 percent increase in pull requests on average, the researchers write. But all that extra code doesn’t translate directly into improved software output on the firm level. On the contrary, the resolution rate for Issues and Epics (i.e. wholesale software features) tracked by tools like Jira did not change in a statistically significant way after AI tools were introduced (the researchers also found no “compositional shift” in the size or complexity of those Jira-tracked issues across the AI introduction).
The reason for that discrepancy can be found directly in the code review process, which takes markedly longer on average after the introduction of AI coding agents. Overall, the average “review process” time between a pull request getting submitted and it being merged into the codebase balloons 49 percent on average after AI agents are introduced. That effect can be seen in more granular data, too, with “the share of pull requests with changes requested nearly doubl[ing], and the number of comments per pull request increas[ing] by 35%” following the AI agent shift, the researchers write.
In response to this change, the researchers found a 14 percent increase in the share of workers performing code reviews after AI agents’ introduction. They also write that they “cannot attribute significant employment changes to AI...”"
https://arstechnica.com/ai/2026/10/ai-coding-agents-generate-more-code-but-not-more-software/
#SoftwareDevelopment #Programming #AI #GenerativeAI #LLMs #AIAgents #Automation #Productivity
Miguel Afonso Caetano 🦾 🌹
npub1ez…apqv4
2026-10-11 11:51:48 UTC
Author Public Key
npub1ezw4ee77tslz4cs0vcrm7gpvdrstgrq2rjpkfrugas33jm3azduqyapqv4Seen on
wss://relay.momostr.pinkPublished at
2026-10-11 11:51:48 UTCEvent JSON
{
"id": "d9123b89e551327533d116560cecf8edd70ddc2b7a19c99b34a2e3774e8c958e",
"pubkey": "c89d5ce7de5c3e2ae20f6607bf202c68e0b40c0a1c83648f88ec23196e3d1378",
"created_at": 1791719508,
"kind": 1,
"tags": [
[
"t",
"programming"
],
[
"t",
"generativeai"
],
[
"t",
"softwaredevelopment"
],
[
"proxy",
"https://tldr.nettime.org/@remixtures/117422129682946067",
"web"
],
[
"t",
"productivity"
],
[
"t",
"llms"
],
[
"t",
"ai"
],
[
"t",
"automation"
],
[
"t",
"aiagents"
],
[
"proxy",
"https://tldr.nettime.org/users/remixtures/statuses/117422129682946067",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://tldr.nettime.org/users/remixtures/statuses/117422129682946067",
"pink.momostr"
],
[
"-"
]
],
"content": "Umm. I think the researchers focused on the wrong thing. For AI productivity benefits to really work when it comes to software development, It's another LLM who should be review most of the code generated by an LLM, while a human should just act mostly as a judge, moderator, and final arbitrer. \n\n\"In terms of raw code being produced, the results are clear and stark. The introduction of AI coding agents at a firm leads to a 30 percent increase in total lines of code generated, a 20 percent rise in the number of total commits, and a 23 percent increase in pull requests on average, the researchers write. But all that extra code doesn’t translate directly into improved software output on the firm level. On the contrary, the resolution rate for Issues and Epics (i.e. wholesale software features) tracked by tools like Jira did not change in a statistically significant way after AI tools were introduced (the researchers also found no “compositional shift” in the size or complexity of those Jira-tracked issues across the AI introduction).\n\nThe reason for that discrepancy can be found directly in the code review process, which takes markedly longer on average after the introduction of AI coding agents. Overall, the average “review process” time between a pull request getting submitted and it being merged into the codebase balloons 49 percent on average after AI agents are introduced. That effect can be seen in more granular data, too, with “the share of pull requests with changes requested nearly doubl[ing], and the number of comments per pull request increas[ing] by 35%” following the AI agent shift, the researchers write.\n\nIn response to this change, the researchers found a 14 percent increase in the share of workers performing code reviews after AI agents’ introduction. They also write that they “cannot attribute significant employment changes to AI...”\"\n\nhttps://arstechnica.com/ai/2026/10/ai-coding-agents-generate-more-code-but-not-more-software/\n\n#SoftwareDevelopment #Programming #AI #GenerativeAI #LLMs #AIAgents #Automation #Productivity",
"sig": "a8b355bd6458fada6e31e5504f9d38a75f2822eb8bae81531c905eb5f0130421bbbbe564429f4ba040e4640b2c26254d4d0782c1535c386e85bf18e777c73025"
}