Large language models (LLMs) have shown potential in supporting decision-making applications, particularly as personal assistants in the financial, healthcare, and legal domains. While prompt engineering strategies have enhanced the capabilities of LLMs in decision-making, cognitive biases inherent to LLMs present significant challenges. Cognitive biases are systematic patterns of deviation from norms or rationality in decision-making that can lead to the production of inaccurate outputs. Existing cognitive bias mitigation strategies assume that input prompts only contain one type of cognitive bias, limiting their effectiveness in more challenging scenarios involving multiple cognitive biases. To fill this gap, we propose a cognitive debiasing approach, self-adaptive cognitive debiasing (SACD), that enhances the reliability of LLMs by iteratively refining prompts. Our method follows three sequential steps - bias determination, bias analysis, and cognitive debiasing - to iteratively mitigate potential cognitive biases in prompts. We evaluate SACD on finance, healthcare, and legal decision-making tasks using both open-weight and closed-weight LLMs. Compared to advanced prompt engineering methods and existing cognitive debiasing techniques, SACD achieves the lowest average bias scores in both single-bias and multi-bias settings.

Large language models (LLMs) offer significant potential as tools to support an expanding range of decision-making tasks. However, given their training on human (created) data, LLMs can inherit both societal biases against protected groups, as well as be subject to cognitive bias. Such human-like bias can impede fair and explainable decisions made with LLM assistance. Our work introduces BiasBuster, a framework designed to uncover, evaluate, and mitigate cognitive bias in LLMs, particularly in high-stakes decision-making tasks. Inspired by prior research in psychology and cognitive sciences, we develop a dataset containing 16,800 prompts to evaluate different cognitive biases (e.g., prompt-induced, sequential, inherent). We test various bias mitigation strategies, amidst proposing a novel method using LLMs to debias their own prompts. Our analysis provides a comprehensive picture on the presence and effects of cognitive bias across different commercial and open-source models. We demonstrate that our self-help debiasing effectively mitigate cognitive bias without having to manually craft examples for each bias type.

Late one recent night, I enlisted GPT-4 to fix my life. I began by soliciting broad-strokes summaries of my journalistic interests and an expedited five-step protocol for breaking in raw-denim jeans (if you know, you know). But after a few rounds, the asks became personal: "How can I tell if I'm overinvested in my career?" and "How do I sum up the volume of my work?" Before I knew it, I'd dredged my reserves of ennui into the early-morning hours, imploring the AI to supply me with maybe-consequential ways of doing--of becoming--better. The answers were sensible enough, delivered in the stiffly efficient prose of a try-hard MBA student--if not quite visionary, just fine.

Pat Calhoun, a visionary leader focused on UX and adoption, is the CEO and Founder of Espressive, transforming enterprise self-help with AI. One of the most dramatic workplace shifts caused by the pandemic is the escalation of digital transformation initiatives. The numbers say it all. According to research by Twilio, 79% of digital transformation budgets grew in response to the pandemic -- and 26% grew "dramatically." Gartner, Inc. also found that over 80% of CEOs have a digital transformation program underway, and 69% are using Covid-19 as a catalyst to focus on resigning their businesses.

More than 90% of surveyed trusts say they still rely on hand-written reports - Photo credit: Flickr, ad.mak, CC BY 2.0 According to figures released following a Freedom of Information request by the communications solutions company Nuance, 93% of the 30 trusts that responded said they handwrite reports. The same proportion said they relied on traditional word processing methods to type up electronic patient records. However, the FoI also indicated an increasing interest in AI across the NHS, with 43% of trusts reporting that they were considering how to use AI as a way of allowing patients to "self-help" when accessing health services. In a statement published alongside the FoI responses, Nuance said that such technology included virtual assistants, speech recognition technology and chat-bots. Last year, Enfield Council became one of the first local authorities to use AI assistance, with the introduction of IPSoft's Amelia systems to help residents carry out online tasks.

One in three NHS trusts are using AI for patient services. Business and government are on a mission to deploy and succeed with AI in healthcare – IBM Watson has been drafted in the fight against cancer, while the NHS is trialling an AI app and working with the likes of Google's Deepmind to improve patient care. It is easy to see why the UK's NHS would be rushing to adopt AI, with the over-burdened, costly and complex infrastructure of the NHS a perfect candidate in which to reap the efficiency and productivity benefits of AI. Although archaic processes still reign supreme in the NHS, new data does suggest that AI is starting to infiltrate and make an impact in trusts around the UK. According to a Freedom of Information request filed by Nuance, one in three NHS trusts are using AI for patient services.

"Smarter Faster Better: The Secrets of Being Productive in Life and Business" (Random House) is Charles Duhigg's follow-up to his best-selling "The Power of Habit: Why We Do What We Do in Life and Business," which was published in 2012. The new book, like its predecessor, has a format that's familiar in contemporary nonfiction: exemplary tales interpolated with a little social and cognitive science. The purpose of the tales is to create entertaining human-interest narratives; the purpose of the science is to help the author pick out a replicable feature of those narratives for readers to emulate. What enabled the pilot to land the badly damaged plane? How did the academic dropout with anxiety disorder become a champion poker player? What made "West Side Story" and Disney's "Frozen" into mega-hits?