Nils Köbis

1️⃣ Big picture: Trust stays ~the same whether you write alone or with AI; and disclosure didn’t meaningfully change that.

2️⃣ The setup: the predictive-text writing assistant participants used (think Smart Compose / Copilot-style suggestions).

3️⃣ The key result: writing with AI didn’t move trust much, but it boosted efficiency: people got more trust per minute spent writing, even when AI use was disclosed.
Takeaway: In one-shot, transactional interactions, AI writing support can increase “trust per time” without clear trust penalties.

This project was led by @zoepurcell.bsky.social, in cooperation with Anne-Marie Nussberger, Mengchen Dong ( @mpib-berlin.bsky.social, CHM) and Maurice Maurice Jakesch (Bauhaus-Universität Weimar).

#AI #Trust #HumanComputerInteraction #BehavioralScience #LLMs #OpenScience #iScience #Transparency

Whether that generalizes to affect-rich or long-term relationships remains to be scrutinized, but we hope our results serve as a first benchmark.
📄 #openaccess Paper:https://www.cell.com/iscience/fulltext/S2589-0042(25)02353-3
📦 Prereg + data + code (OSF): osf.io/nrwb8/

Many people worry that AI-assisted writing feels “less authentic” and may erode trust.
So we tested it in two preregistered, incentivized trust-game experiments (N = 1,637), which just got published in iScience (Cell Press)
More details 👇

People are more likely to cheat when they delegate tasks to AI Artificial intelligence is moving from being a mere tool to an active partner in decision-making, but when we offload misconduct to machines, who bears the blame?

And LLMs are more likely than a human intermediary to comply with prompts that promote cheating

go.nature.com/3VnGUpG

We are on the cover of @nature.com!
Academic dream come true!🤩

People Are More Likely to Cheat When They Use AI Participants in a new study were more likely to cheat when delegating to AI—especially if they could encourage machines to break rules without explicitly asking for it

Now out in Scientific American. Great interview with @nckobis.bsky.social & Zoe Rahwan about our recent @nature.com article.

People Are More Likely to Cheat When They Use AI

www.scientificamerican.com/article/peop...

Thanks @rachelnuwer.bsky.social & @parshallison.bsky.social

Thanks for covering our work @elpais.com

Delegation to artificial intelligence can increase dishonest behaviour - Nature People cheat more when they delegate tasks to artificial intelligence, and large language models are more likely than humans to comply with unethical instructions—a risk that can be minimized by introducing prohibitive, task-specific guardrails.

Nature research paper: Delegation to artificial intelligence can increase dishonest behaviour

go.nature.com/3KsDgbG

Delegation to artificial intelligence can increase dishonest behaviour - Nature People cheat more when they delegate tasks to artificial intelligence, and large language models are more likely than humans to comply with unethical instructions—a risk that can be minimized by ...

Would you let AI (LLMs) cheat for you? New work out in @nature.com shows that people are indeed willing to instruct AI in ways that will benefit themselves, despite not being totally honest. Great work by @nckobis.bsky.social and @iyadrahwan.bsky.social et al 🧪 www.nature.com/articles/s41...

Would you let AI cheat for you?

Our new paper in @nature.com, 5 years in the making, is out today.

www.nature.com/articles/s41...

Delegation to artificial intelligence can increase dishonest behaviour - Nature People cheat more when they delegate tasks to artificial intelligence, and large language models are more likely than humans to comply with unethical instructions—a risk that can be minimized by ...

🚨 Beyond excited that our new paper is out today in @natureportfolio.nature.com

We find that people are more likely to cheat when they delegate to AI than when acting themselves.

www.nature.com/articles/s41...

Is Less More? Field Evidence on the Impact of Anti‐Bribery Policies on Employee Knowledge and Corrupt Behavior Companies increasingly adopt internal norms to enhance compliance with legal rules. However, the rapid growth in volume and complexity of such internal rules may obstruct employee knowledge and under...

💭 Can shorter policies prevent corruption?
In a field experiment, we tested if concise, infographic anti-bribery policies improve employee rule knowledge & reduce corruption.
Spoiler: They didn’t. 🚫
-> social norms turned out to be the real driver.
open-access paper here: doi.org/10.1111/rego...

Workshop opportunity this May in Berlin
With @celbaek.bsky.social, Janis Zickfeld and Zoe Rahwan, we are organizing an EASP Group Meeting on The Future of Behavioral Ethics: Social Perspectives on Dishonesty @easpinfo.bsky.social
More info: www.easp.eu/news/itm/eas...
Deadline: 31st of Jan

Interested in social perspectives on dishonesty?

We are organising a EASP group meeting on the future of behavioral ethics, in Berlin in May.

We invite researchers of all career stages to present their work on dishonesty.

You can find all info here:
www.easp.eu/news/itm/eas...