AI Research
AI that thinks like us—and could help explain how we think
Researchers at Helmholtz Munich have developed an artificial intelligence model that can simulate human behavior with remarkable accuracy. The language model, called Centaur, was trained on more than ten million decisions from psychological experiments—and makes decisions in ways that closely resemble those of real people. This opens new avenues for understanding human cognition and improving psychological theories.
For decades, psychology has aspired to explain the full complexity of human thought. Yet traditional models could either offer a transparent explanation of how people think—or reliably predict how they behave. Achieving both has long seemed out of reach.
The team led by Dr. Marcel Binz and Dr. Eric Schulz, both researchers at the Institute for Human-Centered AI at Helmholtz Munich, has now developed a model that combines both. Centaur was trained using a specially curated dataset called Psych-101, which includes over ten million individual decisions from 160 behavioral experiments. The study is published in the journal Nature.
What makes Centaur unique is its ability to predict human behavior not only in familiar tasks, but also in entirely new situations it has never encountered before. It identifies common decision-making strategies, adapts flexibly to changing contexts—and even predicts reaction times with surprising precision.
“We’ve created a tool that allows us to predict human behavior in any situation described in natural language—like a virtual laboratory,” says Binz, who is also the study’s lead author.
Potential applications range from analyzing classic psychological experiments to simulating individual decision-making processes in clinical contexts—for example, in depression or anxiety disorders. The model opens up new perspectives in health research in particular—for example, by helping us understand how people with different psychological conditions make decisions. The dataset is set to be expanded to include demographic and psychological characteristics.
Centaur: Bridging theory and prediction
Centaur bridges two previously separate domains: interpretable theories and predictive power. It can reveal where classical models fall short—and provide insights into how they might be improved. This opens up new possibilities for research and real‑world applications, from medicine to environmental science and the social sciences.
“We’re just getting started and already seeing enormous potential,” says institute director Schulz. Ensuring that such systems remain transparent and controllable is key, Binz adds—for example, by using open, locally hosted models that safeguard full data sovereignty.
Next, the researchers aim to take a closer look inside Centaur: Which computational patterns correspond to specific decision‑making processes? Can they be used to infer how people process information—or how decision strategies differ between healthy individuals and those with mental health conditions?
The researchers are convinced: “These models have the potential to fundamentally deepen our understanding of human cognition—provided we use them responsibly.” That this research is taking place at Helmholtz Munich rather than in the development departments of major tech companies is no coincidence.
“We combine AI research with psychological theory—and with a clear ethical commitment,” says Binz. “In a public research environment, we have the freedom to pursue fundamental cognitive questions that are often not the focus in industry.”
More information:
Marcel Binz, A foundation model to predict and capture human cognition, Nature (2025). DOI: 10.1038/s41586-025-09215-4. www.nature.com/articles/s41586-025-09215-4
Citation:
Centaur: AI that thinks like us—and could help explain how we think (2025, July 2)
retrieved 4 July 2025
from https://techxplore.com/news/2025-07-centaur-ai.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
AI Research
Researchers Use Hidden AI Prompts to Influence Peer Reviews: A Bold New Era or Ethical Quandary?
AI Secrets in Peer Reviews Uncovered
Last updated:
Edited By
Mackenzie Ferguson
AI Tools Researcher & Implementation Consultant
In a controversial yet intriguing move, researchers have begun using hidden AI prompts to potentially sway the outcomes of peer reviews. This cutting-edge approach aims to enhance review processes, but it raises ethical concerns. Join us as we delve into the implications of AI-assisted peer review tactics and how it might shape the future of academic research.
Introduction to AI in Peer Review
Artificial Intelligence (AI) is rapidly transforming various facets of academia, and one of the most intriguing applications is its integration into the peer review process. At the heart of this evolution is the potential for AI to streamline the evaluation of scholarly articles, which traditionally relies heavily on human expertise and can be subject to biases. Researchers are actively exploring ways to harness AI not just to automate mundane tasks but to provide deep, insightful evaluations that complement human judgment.
The adoption of AI in peer review promises to revolutionize the speed and efficiency with which academic papers are vetted and published. This technological shift is driven by the need to handle an ever-increasing volume of submissions while maintaining high standards of quality. Notably, hidden AI prompts, as discussed in recent studies, can subtly influence reviewers’ decisions, potentially standardizing and enhancing the objectivity of reviews (source).
Incorporating AI into peer review isn’t without challenges. Ethical concerns about transparency, bias, and accountability arise when machines play an integral role in shaping academic discourse. Nonetheless, the potential benefits appear to outweigh the risks, with AI offering tools that can uncover hidden biases and provide more balanced reviews. As described in TechCrunch’s exploration of this topic, there’s an ongoing dialogue about the best practices for integrating AI into these critical processes (source).
Influence of AI in Academic Publishing
The advent of artificial intelligence (AI) is reshaping various sectors, with academic publishing being no exception. The integration of AI tools in academic publishing has significantly streamlined the peer review process, making it more efficient and less biased. According to an article from TechCrunch, researchers are actively exploring ways to integrate AI prompts within the peer review process to subtly guide reviewers’ evaluations without overt influence (). These AI systems analyze vast amounts of data to provide insightful suggestions, thus enhancing the quality of published research.
Moreover, AI applications in academic publishing extend beyond peer review management. AI algorithms can analyze and summarize large datasets, providing researchers with new insights and enabling faster discoveries. As TechCrunch suggests, these technologies are becoming integral to helping researchers manage the ever-increasing volume of scientific literature (). The future of academic publishing might see AI serving as co-authors, providing accurate data analysis and generating hypotheses based on trends across studies.
Public reactions to the influence of AI in academic publishing are mixed. Some view it as a revolutionary tool that democratizes knowledge production by reducing human errors and biases. Others, however, raise concerns over ethical implications, fearing that AI could introduce new biases or be manipulated to favor particular agendas. As TechCrunch highlights, the key challenge will be to implement transparent AI systems that can be held accountable and ensure ethical standards in academic publishing ().
Looking ahead, the influence of AI in academic publishing is poised to grow, potentially transforming various aspects of research dissemination. AI-powered platforms could revolutionize the accessibility and dissemination of knowledge by automating the proofreading and formatting processes, making academic work more readily available and understandable globally. However, as TechCrunch notes, the future implications of such developments require careful consideration to balance innovation with ethical integrity, especially in how AI technologies are governed ().
Challenges and Concerns in AI Implementation
Implementing AI technologies across various sectors presents numerous challenges and concerns, particularly regarding transparency, ethics, and reliability. As researchers strive to integrate AI into processes like peer review, hidden AI prompts can sometimes influence decisions subtly. According to “TechCrunch” in their article about researchers influencing peer review processes with hidden AI prompts, such practices raise questions about the integrity of AI systems . Ensuring AI operates within ethical boundaries becomes crucial, as we must balance innovation with maintaining trust in automated systems.
Furthermore, the opacity of AI algorithms often leads to public and expert concerns about accountability. When AI systems make decisions without clear explanations, it can diminish users’ trust. In exploring the future implications of AI in peer review settings, it becomes apparent that refinements are needed to enhance transparency and ethical considerations. As noted in the TechCrunch article, there is an ongoing debate about the extent to which AI should be allowed to influence decisions that have traditionally been human-centric . This calls for a framework that sets clear standards and guidelines for AI implementation, ensuring its role supplements rather than overrides human judgment.
In addition to transparency and ethics, reliability is another significant concern when implementing AI. The technological robustness of AI systems is continuously tested by real-world applications. Errors or biases in AI can lead to unintended consequences that may affect public perception and acceptance of AI-driven tools. As industries increasingly rely on AI, aligning these systems with societal values and ensuring they are error-free is paramount to gaining widespread acceptance. The TechCrunch article also highlights these reliability issues, suggesting that developers need to focus more on creating accurate, unbiased algorithms .
Experts Weigh in on AI-driven Peer Review
In recent years, the academic community has seen a growing interest in integrating artificial intelligence into the peer review process. Experts believe that AI can significantly enhance this critical phase of academic publishing by bringing in efficiency, consistency, and unbiased evaluation. According to a report on TechCrunch, researchers are exploring ways to subtly incorporate AI prompts into the peer review mechanism to improve the quality of feedback provided to authors (TechCrunch).
The inclusion of AI in peer review is not without its challenges, though. Experts caution that the deployment of AI-driven tools must be done with significant oversight to prevent any undue influence or bias that may occur from automated processes. They emphasize the importance of transparency in how AI algorithms are used and the nature of data fed into these systems to maintain the integrity of peer review (TechCrunch).
While some scholars welcome AI as a potential ally that can alleviate the workload of human reviewers and provide them with analytical insights, others remain skeptical about its impact on the traditional rigor and human judgment in peer evaluations. The debate continues, with public reactions reflecting a mixture of excitement and cautious optimism about the future potential of AI in scholarly communication (TechCrunch).
Public Reactions to AI Interventions
The public’s reaction to AI interventions, especially in fields such as scientific research and peer review, has been a mix of curiosity and skepticism. On one hand, many appreciate the potential of AI to accelerate advancements and improve efficiencies within the scientific community. However, concerns remain over the transparency and ethics of deploying hidden AI prompts to influence processes that traditionally rely on human expertise and judgment. For instance, a recent article on TechCrunch highlighted researchers’ attempts to integrate these AI-driven techniques in peer review, sparking discussions about the potential biases and ethical implications of such interventions.
Further complicating the public’s perception is the potential for AI to disrupt traditional roles and job functions within these industries. Many individuals within the academic and research sectors fear that an over-reliance on AI could undermine professional expertise and lead to job displacement. Despite these concerns, proponents argue that AI, when used effectively, can provide invaluable support to researchers by handling mundane tasks, thereby allowing humans to focus on more complex problem-solving activities, as noted in the TechCrunch article.
Moreover, the ethical ramifications of using AI in peer review processes have prompted a call for stringent regulations and clearer guidelines. The potential for AI to subtly shape research outcomes without the overt consent or awareness of the human peers involved raises significant ethical questions. Discussions in media outlets like TechCrunch indicate a need for balanced discussions that weigh the benefits of AI-enhancements against the necessity to maintain integrity and trust in academic research.
Future of Peer Review with AI
The future of peer review is poised for transformation as AI technologies continue to advance. Researchers are now exploring how AI can be integrated into the peer review process to enhance efficiency and accuracy. Some suggest that AI could assist in identifying potential conflicts of interest, evaluating the robustness of methodologies, or even suggesting suitable reviewers based on their expertise. For instance, a detailed exploration of this endeavor can be found at TechCrunch, where researchers are making significant strides toward innovative uses of AI in peer review.
The integration of AI in peer review does not come without its challenges and ethical considerations. Concerns have been raised regarding potential biases that AI systems might introduce, the transparency of AI decision-making, and how reliance on AI might impact the peer review landscape. As discussed in recent events, stakeholders are debating the need for guidelines and frameworks to manage these issues effectively.
One potential impact of AI on peer review is the democratization of the process, opening doors for a more diverse range of reviewers who may have been overlooked previously due to geographical or institutional biases. This could result in more diverse viewpoints and a richer peer review process. Additionally, as AI becomes more intertwined with peer review, expert opinions highlight the necessity for continuous monitoring and adjustment of AI tools to ensure they meet the ethical standards of academic publishing. This evolution in the peer review process invites us to envision a future where AI and human expertise work collaboratively, enhancing the quality and credibility of academic publications.
Public reactions to the integration of AI in peer review are mixed. Some welcome it as a necessary evolution that could address long-standing inefficiencies in the system, while others worry about the potential loss of human oversight and judgment. Future implications suggest a field where AI-driven processes could eventually lead to a more streamlined and transparent peer review system, provided that ethical guidelines are strictly adhered to and biases are meticulously managed.
An Xbox producer has faced a backlash after suggesting laid-off employees should use artificial intelligence to deal with emotions in a now deleted LinkedIn post.
Matt Turnbull, an executive producer at Xbox Game Studios Publishing, wrote the post after Microsoft confirmed it would lay off up to 9,000 workers, in a wave of job cuts this year.
The post, which was captured in a screenshot by tech news site Aftermath, shows Mr Turnbull suggesting tools like ChatGPT or Copilot to “help reduce the emotional and cognitive load that comes with job loss.”
One X user called it “plain disgusting” while another said it left them “speechless”. The BBC has contacted Microsoft, which owns Xbox, for comment.
Microsoft previously said several of its divisions would be affected without specifying which ones but reports suggest that its Xbox video gaming unit will be hit.
Microsoft has set out plans to invest heavily in artificial intelligence (AI), and is spending $80bn (£68.6bn) in huge data centres to train AI models.
Mr Turnbull acknowledged the difficulty of job cuts in his post and said “if you’re navigating a layoff or even quietly preparing for one, you’re not alone and you don’t have to go it alone”.
He wrote that he was aware AI tools can cause “strong feelings in people” but wanted to try and offer the “best advice” under the circumstances.
The Xbox producer said he’d been “experimenting with ways to use LLM Al tools” and suggested some prompts to enter into AI software.
These included career planning prompts, resume and LinkedIn help, and questions to ask for advice on emotional clarity and confidence.
“If this helps, feel free to share with others in your network,” he wrote.
The Microsoft cuts would equate to 4% of Microsoft’s 228,000-strong global workforce.
Some video game projects have reportedly been affected by the cuts.
For internationally operating companies, it is attractive to use a single AI solution across all markets. Such a centralized approach offers economies of scale and appears to ensure uniformity. Yet research from CWI reveals that this assumption is on shaky ground: the language in which an AI is addressed, influences the answers the system provides – and quite significantly too.
Language steers outcomes
The problem goes beyond small differences in nuance. Researcher Davide Ceolin, tenured researcher within the Human-Centered Data Analytics group at CWI, and his international research team discovered that identical Large Language Models (LLM’s) can adopt varying political standpoints, depending on the language used. They delivered more economically progressive responses in Dutch and more centre-conservative ones in English. For organizations applying AI in HR, customer service or strategic decision-making, this results in direct consequences for business processes and reputation.
These differences are not incidental. Statistical analysis shows that the language of the prompt used has a stronger influence on the AI response than other factors, such as assigned nationality. “We assumed that the output of an AI model would remain consistent, regardless of the language. But that turns out not to be the case,” says Ceolin.
For businesses, this means more than academic curiosity. Ceolin emphasizes: “When a system responds differently to users with different languages or cultural backgrounds, this can be advantageous – think of personalization – but also detrimental, such as with prejudices. When the owners of these systems are unaware of this bias, they may experience harmful consequences.”
Prejudices with consequences
The implications of these findings extend beyond political standpoints alone. Every domain in which AI is deployed – from HR and customer service to risk assessment – runs the risk of skewed outcomes as a result of language-specific prejudices. An AI assistant that assesses job applicants differently depending on the language of their CV, or a chatbot that gives inconsistent answers to customers in different languages: these are realistic scenarios, no longer hypothetical.
According to Ceolin, such deviations are not random outliers, but patterns with a systematic character. “That is extra concerning. Especially when organizations are unaware of this.”
For Dutch multinationals, this is a real risk. They often operate in multiple languages but utilize a single central AI system. “I suspect this problem already occurs within organizations, but it’s unclear to what extent people are aware of it,” says Ceolin. The research also suggests that smaller models are, on average, more consistent than the larger, more advanced variants, which appear to be more sensitive to cultural and linguistic nuances.
What can organizations do?
The good news is that the problem can be detected and limited. Ceolin advises testing AI systems regularly using persona-based prompting, which involves testing different scenarios where the language, nationality, or culture of the user varies. “This way you can analyze whether specific characteristics lead to unexpected or unwanted behaviour.”
Additionally, it’s essential to have a clear understanding of who works with the system and in which language. Only then you can assess whether the system operates consistently and fairly in practice. Ceolin advocates for clear governance frameworks that account for language-sensitive bias, just as currently happens with security or ethics.
Structural approach required
According to the researchers, multilingual AI bias is not a temporary phenomenon that will disappear on its own. “Compare it to the early years of internet security,” says Ceolin. “What was then seen as a side issue turned out to be of strategic importance later.” CWI is now collaborating with the French partner institute INRIA to unravel the mechanisms behind this problem further.
The conclusion is clear: companies that deploy AI in multilingual contexts would do well to consciously address this risk not only for technical reasons, but also to prevent reputational damage, legal complications and unfair treatment of customers or employees.
“AI is being deployed increasingly often, but insight into how language influences the system is in its infancy,” concludes Ceolin. “There’s still much work to be done there.”
Author: Kim Loohuis
Header photo: Shutterstock
-
Funding & Business7 days agoKayak and Expedia race to build AI travel agents that turn social posts into itineraries
-
Jobs & Careers6 days ago
Mumbai-based Perplexity Alternative Has 60k+ Users Without Funding
-
Mergers & Acquisitions6 days ago
Donald Trump suggests US government review subsidies to Elon Musk’s companies
-
Funding & Business6 days ago
Rethinking Venture Capital’s Talent Pipeline
-
Jobs & Careers6 days ago
Why Agentic AI Isn’t Pure Hype (And What Skeptics Aren’t Seeing Yet)
-
Funding & Business4 days ago
Sakana AI’s TreeQuest: Deploy multi-model teams that outperform individual LLMs by 30%
-
Jobs & Careers6 days ago
Astrophel Aerospace Raises ₹6.84 Crore to Build Reusable Launch Vehicle
-
Funding & Business7 days agoFrom chatbots to collaborators: How AI agents are reshaping enterprise work
-
Tools & Platforms6 days ago
Winning with AI – A Playbook for Pest Control Business Leaders to Drive Growth
-
Jobs & Careers4 days ago
Ilya Sutskever Takes Over as CEO of Safe Superintelligence After Daniel Gross’s Exit
