chatgpt Timeline - FediScanner

For years economists urged calm about technology‑driven job fears, pointing to ATMs, Excel and robotic vacuums as examples of tools that augment rather than replace workers. That consensus is now cracking. A new paper by researchers at the Federal Reserve Bank of Chicago, the Forecasting Research Institute, Yale, Stanford and the University of Pennsylvania surveyed 69 economists, 52 AI specialists and 38 superforecasters on AI’s impact on the U.S. economy. All three groups agree that faster AI progress will depress labor‑force participation – the polite way of saying “fewer people working.” In the “rapid” scenario, where AI eclipses human performance across most tasks by 2030, economists project the U.S. labor‑force participation rate falling from the current 62 % to 54 % by 2050, with roughly 10 million jobs lost directly to AI rather than demographics. The same scenario forecasts annual GDP growth rising to about 3.5 % by 2045‑2049, while AI experts are even more bullish at 5.3 %, creating tremendous aggregate wealth that would become heavily concentrated at the top—potentially 80 % of total wealth held by the richest 10 % of households by 2050.

The paper highlights a shift in the debate: disagreement among experts is no longer about whether powerful AI will arrive, but about its economic consequences. Earlier pro‑tech arguments assumed automation would eventually generate new categories of work, but the new question is whether AI can automate the very act of inventing new tasks. Recent data show a mixed picture. A Yale‑Brookings study from late 2025 found no mass‑unemployment signal three years after ChatGPT’s launch, yet the same research documents a 13 % relative employment drop for workers aged 22‑25 in the most AI‑exposed occupations. The macro‑level employment figures look stable, but the leading edge of the labor market is already feeling the strain.

#federalreserve #forecastingresearch #yale #chatgpt #u.s.

Show Original Post

Report

12.04.2026 09:22
habr (@habr@zhub.link)

Почему тексты нейросетей раздражают читателя и как я с ними работаю

Почему тексты нейросетей часто вызывают раздражение даже без явных ошибок. Разбираю типичные проблемы ИИ-контента и расскажу, как я правлю такие тексты, чтобы их можно было нормально читать.

https://habr.com/ru/articles/1022426/

#нейросети #тексты #копирайтинг #ai #редактура #контент #seo #генерация_текста #chatgpt #статьи

Show Original Post

Report

12.04.2026 08:40
kenjiai_tips (@kenjiai_tips@mastodon.social)

「紹介営業」をChatGPTで加速させる方法。既存顧客に紹介してもらいやすい瞬間がある。導入後1〜2ヶ月の成果が出始めた直後が最適タイミング。お礼メール＋紹介依頼を同時に送る文面テンプレートをChatGPTで作ったら紹介率が上がった。 #ChatGPT #AI活用 #営業

Show Original Post

Report

12.04.2026 08:33
karlos (@karlos@mindly.social)

@lambda I agree about NOT burning Earth by using AI. It is a last option for me. The next generation uses AI just as a Tool. And I hope they mind, AI is as erring as the human brain. We will have enough energy for AI - some day. But now, as people still vote for capitalism, there is no government to control it, to only let the data center be build, if it is powered by renewables.

For the Java-to-Rust, it was ment as a one shot. Transcode the Java files, and done. But it came out different. AI "likes" to please humans. It seemed to "know" I like quick running results and offered a minimised solution. Just read and analyse the XML OSM tagging, get the building footprints and write a dummy gltf file.

Not surprisingly, there were Rust errors like missing lifetimes. Surprisingly, I could "tell" them to #ChatGPT and it found a solution. And yes, one should study the solutions to learn something. More logical errors were not found, like splitting the gltf buffer in parts for positions etc. Reading and minding the code myself was needed.

And still, the gltf had errors, even I do not understand. Sure, I could study gltf. But as the Java code is ok, This way isn't an approbate solution.
Also the small steps, the AI offered, to implement the whole function of OSM2World would take weeks; and a hell of energy, mental an physical.

Why at all AI? Transcoding should be able by a brain-made application. Prompt: write an app for it. ;-)

Show Original Post

Report

12.04.2026 07:40
kenjiai_tips (@kenjiai_tips@mastodon.social)

AIを使っている営業マンと使っていない営業マンの差が、来年には取り返せないレベルになると思っている。魔法みたいには変わらないけど、毎日の準備・振り返り・改善が少し速く少し精度が高くなる。その積み上げが半年後に大きな差になる。 #ChatGPT #AI活用 #営業

Show Original Post

Report