AI Expansion Brings Cost Savings, Rising Power Demand and New Safety Debates
publishers 36articles 57first reported 23 Sep, 14:00 UTCdeveloping since 30 Sep · 32 editions
Artificial intelligence adoption is accelerating across businesses and research organizations, while companies and policymakers confront uncertain returns, mounting electricity demand and disagreements over safety.
Booking Holdings CFO Ewout Steenbergen said at Fortune’s AIQ Summit that even large technology companies investing hundreds of billions of dollars in AI may not know what returns they will generate. At Booking, referrals from large language models accounted for less than 1% of room nights, but internal uses have produced clearer gains. Steenbergen said customer-service costs declined slightly despite high-single-digit booking growth, while about 9,000 engineers increased the amount of tested code reaching production by roughly 30%.
The expansion is also increasing pressure on the power grid. CBS News reported that Schneider Electric projects U.S. electricity demand will rise 16% by 2029, driven mainly by data centers. Data centers consumed about 4.4% of U.S. electricity in 2023, according to Lawrence Berkeley National Laboratory. Consumer advocates warn that households may bear infrastructure costs through higher utility rates.
Safety researchers remain divided over whether advanced AI development should be open or restricted. WIRED reported that researchers Nathan Lambert and Tom Zick founded the nonprofit Trillium Labs to publish experiments involving AI agents and potential recursive self-improvement for outside scrutiny and replication.
Separately, former OpenAI safety lead David Robinson argued that frontier-model releases should follow practices used by nuclear plants and busy airports, with redundant safeguards and deliberate planning. Engadget reported that Robinson criticized OpenAI’s launch culture after leaving the company and warned that current protections lack sufficient rigor.
HOW THIS STORY WAS MADE
Written from 8 articles, 6 with the publisher's own text; 8 independent newsrooms once syndicated copies count as one; this version written 68 h after the record first saw the story; the editor kept it as written.
- Publishers
- 36
- Source articles
- 57
- Given to the writer
- 8, 6 with the article's own text
- Independent newsrooms
- 8
- This version written
- 68 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →