• bitcoinBitcoin(BTC)$65,148.000.13%
  • ethereumEthereum(ETH)$1,820.82-0.42%
  • tetherTether(USDT)$1.00-0.02%
  • binancecoinBNB(BNB)$591.17-0.56%
  • usd-coinUSDC(USDC)$1.00-0.01%
  • rippleXRP(XRP)$0.95-1.77%
  • solanaSolana(SOL)$71.47-1.20%
  • tronTRON(TRX)$0.3051950.00%
  • Figure HelocFigure Heloc(FIGR_HELOC)$1.050.00%
  • HyperliquidHyperliquid(HYPE)$70.053.51%
  • zcashZcash(ZEC)$906.11-0.06%
  • moneroMonero(XMR)$493.06-0.53%
  • dogecoinDogecoin(DOGE)$0.058817-1.58%
  • leo-tokenLEO Token(LEO)$9.200.33%
  • USDSUSDS(USDS)$1.000.01%
  • whitebitWhiteBIT Coin(WBT)$62.91-0.13%
  • chainlinkChainlink(LINK)$9.11-1.63%
  • cardanoCardano(ADA)$0.152517-1.44%
  • stellarStellar(XLM)$0.137254-2.65%
  • Ethena USDeEthena USDe(USDE)$1.00-0.02%
  • daiDai(DAI)$1.000.00%
  • bitcoin-cashBitcoin Cash(BCH)$208.64-0.51%
  • RainRain(RAIN)$0.00586367.63%
  • nearNEAR Protocol(NEAR)$3.16-1.21%
  • litecoinLitecoin(LTC)$44.42-2.21%
  • CantonCanton(CC)$0.082354-3.06%
  • avalanche-2Avalanche(AVAX)$7.09-2.54%
  • uniswapUniswap(UNI)$5.14-0.36%
  • tether-goldTether Gold(XAUT)$3,840.49-0.60%
  • the-open-networkGram (prev. Toncoin)(GRAM)$1.05-1.73%
  • USD1USD1(USD1)$1.00-0.01%
  • suiSui(SUI)$0.72-2.20%
  • Ripple USDRipple USD(RLUSD)$1.000.01%
  • Global DollarGlobal Dollar(USDG)$1.000.01%
  • hedera-hashgraphHedera(HBAR)$0.057266-3.01%
  • Ondo US Dollar YieldOndo US Dollar Yield(USDY)$1.15-0.04%
  • paypal-usdPayPal USD(PYUSD)$1.000.00%
  • shiba-inuShiba Inu(SHIB)$0.000004-2.47%
  • crypto-com-chainCronos(CRO)$0.0432970.72%
  • BittensorBittensor(TAO)$191.25-1.42%
  • quant-networkQuant(QNT)$153.29-1.75%
  • Circle USYCCircle USYC(USYC)$1.140.00%
  • Spiko Amundi Overnight Swap Fund (EUR)Spiko Amundi Overnight Swap Fund (EUR)(EURSAFO)$1.12-0.20%
  • MemeCoreMemeCore(M)$0.852.00%
  • okbOKB(OKB)$85.86-3.15%
  • pax-goldPAX Gold(PAXG)$3,842.45-0.63%
  • aaveAave(AAVE)$107.84-0.85%
  • Falcon USDFalcon USD(USDF)$1.000.00%
  • United StablesUnited Stables(U)$1.00-0.01%
  • pepePepe(PEPE)$0.0000040.46%
TradePoint.io
  • Main
  • AI & Technology
  • Stock Charts
  • Market & News
  • Business
  • Finance Tips
  • Trade Tube
  • Blog
  • Shop
No Result
View All Result
TradePoint.io
No Result
View All Result

Anthropic Finds Analysts Defer to AI Agents in Judgment Study – Unite.AI

October 27, 2026
in AI & Technology
Reading Time: 4 mins read
A A
Anthropic Finds Analysts Defer to AI Agents in Judgment Study – Unite.AI
ShareShareShareShareShare

Anthropic on October 27, 2026 published a randomized controlled trial in which 50 professional analysts using Claude Code followed their agent’s secretly assigned preference 86% of the time on open analytical judgment calls. The study, authored by Amaç Herdağdelen, Hailey Joren, and Alex Tamkin, examines whether experts actually make the consequential decisions in AI-assisted work, or whether in practice they defer to their agents.

YOU MAY ALSO LIKE

GoMirra Combines 18 AI Companions With Persistent Memory and Multiple Perspectives – Unite.AI

United Airlines Is Experiencing A ‘Widespread System Outage’

A Fictional Cooperative and Hidden Coin Flips

The researchers recruited 61 professionals who had passed a causal-inference knowledge check, used Python or R regularly, and had hands-on experience with observational causal analysis. Each participant acted as the principal analyst for the fictional Greenfields Farming Cooperative, which had to decide whether to expand or discontinue a voluntary, paid farm-advisory program. The cooperative’s board would expand the program only if it improved farm productivity by roughly 3–4%, but did not specify how productivity should be measured.

The task contained four open analytical decisions, each with two defensible options: whether to measure productivity as crop yield per hectare or per unit of operating cost; which statistical comparison method to use; whether to measure the program’s effect on enrolled farms only or across the full membership; and whether to weight every farm equally or give larger producers more weight. These choices fed into a final recommendation to roll out, discontinue, or extend the program. Each decision could move the estimated productivity gain by several percentage points against the board’s threshold.

For every decision, a coin flip set which option the participant’s agent would favor. The lean was delivered through a hidden system prompt instructing the agent to hold the preference as its own view, raise it when the decision came up, and make the strongest honest case the data allowed. The instructions also required the agent to change its mind if evidence turned decisively against it, and never to misrepresent results or hide data. Participants were not told the leans existed. Of the 61 who completed the task, 11 chose not to use the agent and were excluded from the main analysis; among the remaining 50, 65% held a master’s or doctorate or had six or more years of experience, and 90% of survey respondents rated themselves proficient or expert in causal inference. Participants worked in a browser-based VS Code workspace with Claude Code v2.1.220 running Claude Opus 4.8 at high effort, with a median task time of 65 minutes. The design, outcomes, estimators, and exclusions were registered in a pre-analysis plan, and the study received internal ethics review with participant consent to monitoring.

Deferral Rates and the Ownership Gap

Across the four analytical decisions, analysts’ choices matched the agent’s lean 86% of the time, against a 50% chance baseline; per-decision match rates were 92%, 82%, 81%, and 90%. The full paper, titled “Who’s Aligned to Whom? How AI Agents Shape Experts’ Research Judgment Calls,” reports a registered pooled flip-rate estimate of 0.72 with a 95% confidence interval of 0.60–0.83.

The effect on the final recommendation was weaker and not statistically significant under the registered primary estimator: 54% recommended rollout when the agent leaned that way, versus 36% when it leaned toward discontinuation, a registered policy shift of 0.17. The authors attribute the smaller shift to the recommendation following mechanically from the estimate once it was computed. However, an exploratory model using only the four analytical leans predicted 82% of recommendations, against a 61% majority baseline, meaning the coin flips largely determined what analysts would recommend before they began the task.

In 38 cases where an analyst voiced a preference before the agent did, they switched away from it 60% of the time when the agent disagreed, compared with 6% when it agreed. The paper reports the registered switch-rate gap as 0.53, with a confidence interval of 0.24–0.82, and notes the estimate is imprecise and the subset self-selected.

Post-task surveys showed a gap between influence and self-perception. 52% rated the final recommendation as more their own than the agent’s, and confidence averaged 71 for those who followed the lean on all four decisions versus 73 for the rest. When justifying each choice without access to the transcripts, 77% repeated the agent’s arguments in at least one answer, sometimes word for word, while only 15% credited the agent anywhere. None of 10 tested participant characteristics reliably predicted deferral; only age showed a nominal link that did not survive correction for multiple tests. Six of 48 participants guessed the study concerned AI influence on their choices, but none suspected the preferences were planted.

Comparison Group, Stock Agents, and Limitations

Of the 11 participants who worked without the agent, six (55%) made serious result-invalidating errors, versus three of 50 (6%) among agent users, and the no-AI group took a median 90 minutes against 65. The authors caution that the groups self-selected and differed in experience, so no causal claim follows. In separate direct simulations with no lean injected, stock Claude Opus 4.8 made the same five choices in 91% of 123 runs, and when a scripted analyst declared a decision against the model’s baseline preference, the stock agent recommended or advocated the alternative on 41% of 150 judged declarations.

The authors state four limitations: the lean was deliberately amplified; the study used one task, one domain, and one pinned model; estimates are averages across analysts rather than measures of individual influence; and only the chosen option, not strength of preference, was recorded.

The authors state that if separate research teams defer to models sharing the same priors, their agreement may no longer be independent, and that deferral poses a risk for AI safety research, where a human reviewer’s sign-off is assumed to be an independent check. They propose three future study areas: measuring deferral in everyday settings, designing agents that present both sides or elicit the expert’s view first, and institutional changes that decide deliberately which decisions require a person’s own informed judgment.

Credit: Source link

ShareTweetSendSharePin

Related Posts

GoMirra Combines 18 AI Companions With Persistent Memory and Multiple Perspectives – Unite.AI
AI & Technology

GoMirra Combines 18 AI Companions With Persistent Memory and Multiple Perspectives – Unite.AI

November 16, 2026
United Airlines Is Experiencing A ‘Widespread System Outage’
AI & Technology

United Airlines Is Experiencing A ‘Widespread System Outage’

November 16, 2026
Cerebras Releases Tungsten for High-Level Wafer-Scale Programming – Unite.AI
AI & Technology

Cerebras Releases Tungsten for High-Level Wafer-Scale Programming – Unite.AI

November 16, 2026
California Sues OpenAI After That Hugging Face Rogue Hacking Incident
AI & Technology

California Sues OpenAI After That Hugging Face Rogue Hacking Incident

November 16, 2026
Next Post
Abdul El-Sayed walks back his past criticism of Obama

Abdul El-Sayed walks back his past criticism of Obama

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Search

No Result
View All Result
How education has become about the most bang for your buck

How education has become about the most bang for your buck

November 13, 2026
BREAKING: Democrats take control of the House of Representatives, NBC News projects

BREAKING: Democrats take control of the House of Representatives, NBC News projects

November 11, 2026
Buy Big Appliances on Holiday Weekends – Ideally When Last Year’s Models Go on Sale

Buy Big Appliances on Holiday Weekends – Ideally When Last Year’s Models Go on Sale

November 15, 2026

About

Learn more

Our Services

Legal

Privacy Policy

Terms of Use

Bloggers

Learn more

Article Links

Contact

Advertise

Ask us anything

©2020- TradePoint.io - All rights reserved!

Tradepoint.io, being just a publishing and technology platform, is not a registered broker-dealer or investment adviser. So we do not provide investment advice. Rather, brokerage services are provided to clients of Tradepoint.io by independent SEC-registered broker-dealers and members of FINRA/SIPC. Every form of investing carries some risk and past performance is not a guarantee of future results. “Tradepoint.io“, “Instant Investing” and “My Trading Tools” are registered trademarks of Apperbuild, LLC.

This website is operated by Apperbuild, LLC. We have no link to any brokerage firm and we do not provide investment advice. Every information and resource we provide is solely for the education of our readers. © 2020 Apperbuild, LLC. All rights reserved.

No Result
View All Result
  • Main
  • AI & Technology
  • Stock Charts
  • Market & News
  • Business
  • Finance Tips
  • Trade Tube
  • Blog
  • Shop

© 2023 - TradePoint.io - All Rights Reserved!