News for hackers. Stuff that matters.

Gillesray

Gillesray on Hacker News

Should Sam Altman fear token compression?

Posted by Gillesray on Thursday March 12, 2026 @12:37PM
from the self-post dept.

AI bills are exploding despite costs dropping 10x every 18 months because every saving gets reinvested into more usage.BUT: - Today's prices are subsidized (OpenAI losing $14B in 2026, Anthropic burning $12B in one quarter, Cursor alleges that a $200/month Claude Code subscription could cost up to $5,000 in compute.) - When subsidies end, prices reverse into a 10x larger consumption base -...

( Read More... | no comments | 6 points | on HN )
Re:Should Sam Altman fear token compression? on Thursday March 12, 2026 @12:35PM (#47353468)
AI bills are exploding despite costs dropping 10x every 18 months because every saving gets reinvested into more usage.

BUT: - Today's prices are subsidized (OpenAI losing $14B in 2026, Anthropic burning $12B in one quarter, Cursor alleges that a $200/month Claude Code subscription could cost up to $5,000 in compute.) - When subsidies end, prices reverse into a 10x larger consumption base - That's the double squeeze nobody is modeling for Companies building token efficiency now won't feel it. The rest will. And Altman? Cheaper AI to run = more use cases justified = bigger market for frontier models.

Re:Who's in Charge? Disempowerment Patterns in Real-World LLM Usage on Monday February 23, 2026 @03:01AM (#47119380)
Although AI assistants are now deeply embedded in society, there has been limited empirical study of how their usage affects human empowerment. We present the first large-scale empirical analysis of disempowerment patterns in real-world AI assistant interactions, analyzing 1.5 million consumer Claude.ai conversations using a privacypreserving approach. We focus on situational disempowerment potential, which occurs when AI assistant interactions risk leading users to form distorted perceptions of reality, make inauthentic value judgments, or act in ways misaligned with their values.
Re:Measuring AI agent autonomy in practice on Monday February 23, 2026 @02:54AM (#47119333)
50% around coding, they nail the first usage
Re:Token Compression, achieving more with less on Thursday February 19, 2026 @10:18AM (#47074720)
very interesting
Re:Will I Be Paid in Tokens? on Wednesday February 18, 2026 @06:07AM (#47059770)
indeed. It tells a lot of about token ecosystem
Re:Will I Be Paid in Tokens? on Wednesday February 18, 2026 @02:00AM (#47058102)
Technology companies are adding a fourth component to engineering compensation : salary, bonus, options, & inference costs. Levels.fyi pegs the 75th percentile software engineer salary at $375k. Add $100k in inference & the fully loaded cost is $475k. That’s 21% in tokens.
Re:AI economics looks like cloud's one on Tuesday February 10, 2026 @04:16AM (#46957169)
Enterprise AI spending jumped 3.2× in 2025 to $37B (Gartner), while inference costs dropped 90%.

The paradox isn't new, it's cloud economics on fast-forward.

What took cloud 10 years to reveal is happening in AI in 2-3 years: a) Low unit costs → elastic, untracked adoption b) Agentic workflows multiply requests 10-50× c) Cost drivers are architectural (retries, context bloat), not volumetric d) Finance sees the bill only when it's too late to re-architect

Rigorous cost governance will be essential to realizing AI ROI.

Re:AlexNet Source Code Released on Thursday March 20, 2025 @01:17PM (#43426012)
great sum up
Re:AlexNet Source Code Released on Thursday March 20, 2025 @01:17PM (#43426011)
love it
Re:AI firms will soon exhaust most of the internet's data on Wednesday August 14, 2024 @10:54PM (#41252766)
"One approach is to focus on data quality rather than quantity. ai labs do not simply train their models on the entire internet. They filter and sequence data to maximise how much their models learn. Naveen Rao of Databricks, an ai firm, says that this is the “main differentiator” between ai models on the market. “True information” about the world obviously matters; so does lots of “reasoning”." said Naveen Rao

Edgee demo day. See how to collect 100% of the data

Posted by Gillesray on Wednesday July 17, 2024 @03:48PM
from the self-post dept.

WOW! tomorrow we are doing our first demo day and there are only 5 places left (or maybe even fewer by the time you read this). If you want to see how much usage data are missing on your analytics, and how to collect 100% of the data. Join us.It takes place tomorrow, July 18, at 5 pm CET.https://www.eventbrite.fr/e/edgee-demo-day-tickets-939243461477If no more places are available, another...

( Read More... | no comments | on HN )

The Big Lie

Posted by Gillesray on Tuesday June 18, 2024 @04:32AM
from the self-post dept.

It started this dayIn early 2023, Sacha, then CTO of Le Monde, France's leading news outlet with 300 millions page views and 46 millions of users monthly, faced a critical issue. Despite his pursuit of perfection, he noticed discrepancies in core metrics reported in corporate dashboards, weekly activity reports, and monthly management reports.Key indicators such as Monthly Active Users (MAU),...

( Read More... | no comments | 3 points | on HN )