Large Language Models (LLMs) like gpt-3.5-turbo and claude-instant-1.2 have become instrumental in interpreting and executing semantic-based tasks. Unfortunately, these models' inherent biases, akin to human cognitive biases, adversely affect their performance. Particularly affected is object selection from lists; a fundamental operation in digital navigation and decision-making. This research critically examines these biases and quantifies the effects on a representative list selection task. To explore these biases, we conducted a series of controlled experiments, manipulating temperature, list length, object identity, object type, prompt complexity, and model. This enabled us to isolate and measure the influence of the biases on selection behavior. Our findings show that bias structure is strongly dependent on the model, with object type modulating the magnitude of the effect. With a strong primacy effect, causing the first objects in a list to be disproprotionately represented in outputs. Furthermore the usage of guard rails, a prompt engineering method of ensuring a response structure, can increase bias and decrease instruction adherence when combined with a selection task. The bias is ablated when the guard rail step is separated from the list sampling step, lowering the complexity of each individual task. The implications of this research are two-fold, practically providing a guide for designing unbiased LLM applications and theoretically suggesting that LLMs experience a form of cognitive load compensated for by increasing bias.

The UK will lead on limiting the dangers of artificial intelligence, Rishi Sunak has said, after calls from some tech experts and business leaders for a moratorium. Sunak said AI could bring benefits and prove transformative for society, but it had to be introduced "safely and securely with guard rails in place". The prime minister's comments sound a more cautious approach than in the past, after tech leaders including Twitter's owner, Elon Musk, and Apple's co-founder Steve Wozniak added their names to nearly 30,000 signatures on a letter urging a pause in significant projects. The letter called for a moratorium while the capabilities and dangers of systems such as ChatGPT-4 are properly studied and mitigated in response to fears about the creation of digital minds, fraud, disinformation and the risk to jobs. Sunak has been an advocate of AI, emphasising its benefits rather than risks, and in March the government unveiled a light-touch regulatory programme that did not appear to include proposals for any new laws or enforcement bodies.

Making AI more'humble' could not only help improve AI decision making, but could also help inspire ... [ ] more trust in the technology as a whole, and open the door for more useful and mission-critical applications in the future. AI is notoriously difficult to explain, and some deep learning algorithms can be too complex for even their creators to understand their reasoning. This makes it hard to trust what AI is doing, and even harder to find mistakes before it's too late. Having an algorithm stop partway through its reasoning to check with a human-in-the-loop could inspire more trust in AI, and open the door for the technology to be used in more sensitive and mission-critical applications. Injecting some'humility' into AI in this way could not only make AI more trustworthy and change how companies think about AI, but it could also help to demystify AI and reveal it as the logical and reliable technology that it is.

Artificial intelligence technology will "create a fundamentally different kind of financial system", says the World Economic Forum, which is flagging big challenges for banks to explain, to customers and regulators, how decisions are being made by systems designed to evolve autonomously. AI will create new vulnerabilities for financial stability stemming from its interconnectedness and the herding instincts of algorithms, which could exacerbate crises, the forum suggests. It points to algorithms "destabilising competition" and possible blowback for banks given the technology could introduce unfair biases, propagating financial disadvantage, and erode human banking skills. But at the same time the forum is guiding financial services players on how to implement AI, which it says could ultimately improve the delivery of financial services to customers. A core focus should be policies to explain how decisions are made. "The enormous complexity of some AI systems makes it difficult to obtain an interpretable explanation for why the system has produced a given output," the report says.