Large language models (LLMs) are widely applied across diverse domains, raising concerns about their limitations and potential risks. In this study, we investigate two types of bias that LLMs may display: stereotype bias and deviation bias. Stereotype bias refers to when LLMs consistently associate specific traits with a particular demographic group. Deviation bias reflects the disparity between the demographic distributions extracted from LLM-generated content and real-world demographic distributions. By asking four advanced LLMs to generate profiles of individuals, we examine the associations between each demographic group and attributes such as political affiliation, religion, and sexual orientation. Our experimental results show that all examined LLMs exhibit both significant stereotype bias and deviation bias towards multiple groups. Our findings uncover the biases that occur when LLMs infer user attributes and shed light on the potential harms of LLM-generated outputs.

We introduce Pixtral-12B, a 12--billion-parameter multimodal language model. Pixtral-12B is trained to understand both natural images and documents, achieving leading performance on various multimodal benchmarks, surpassing a number of larger models. Unlike many open-source models, Pixtral is also a cutting-edge text model for its size, and does not compromise on natural language performance to excel in multimodal tasks. Pixtral uses a new vision encoder trained from scratch, which allows it to ingest images at their natural resolution and aspect ratio. This gives users flexibility on the number of tokens used to process an image. Pixtral is also able to process any number of images in its long context window of 128K tokens. Pixtral 12B substanially outperforms other open models of similar sizes (Llama-3.2 11B \& Qwen-2-VL 7B). It also outperforms much larger open models like Llama-3.2 90B while being 7x smaller. We further contribute an open-source benchmark, MM-MT-Bench, for evaluating vision-language models in practical scenarios, and provide detailed analysis and code for standardized evaluation protocols for multimodal LLMs. Pixtral-12B is released under Apache 2.0 license.

Economic issues, such as inflation, energy costs, taxes, and interest rates, are a constant presence in our daily lives and have been exacerbated by global events such as pandemics, environmental disasters, and wars. A sustained history of financial crises reveals significant weaknesses and vulnerabilities in the foundations of modern economies. Another significant issue currently is people quitting their jobs in large numbers. Moreover, many organizations have a diverse workforce comprising multiple generations posing new challenges. Transformative approaches in economics and labour markets are needed to protect our societies, economies, and planet. In this work, we use big data and machine learning methods to discover multi-perspective parameters for multi-generational labour markets. The parameters for the academic perspective are discovered using 35,000 article abstracts from the Web of Science for the period 1958-2022 and for the professionals' perspective using 57,000 LinkedIn posts from 2022. We discover a total of 28 parameters and categorised them into 5 macro-parameters, Learning & Skills, Employment Sectors, Consumer Industries, Learning & Employment Issues, and Generations-specific Issues. A complete machine learning software tool is developed for data-driven parameter discovery. A variety of quantitative and visualisation methods are applied and multiple taxonomies are extracted to explore multi-generational labour markets. A knowledge structure and literature review of multi-generational labour markets using over 100 research articles is provided. It is expected that this work will enhance the theory and practice of AI-based methods for knowledge discovery and system parameter discovery to develop autonomous capabilities and systems and promote novel approaches to labour economics and markets, leading to the development of sustainable societies and economies.

Artificial intelligence is everywhere in our day to day lives. But, with the world's media filled with dystopian tales of computer intelligence taking over humanity; what does the public really think of AI? Using data1 from Google Search Trends, Linkfluence, and Answer the Public, Ebuyer has revealed that over 2.9 million negative posts were generated on AI over the past year! So, does the public really trust AI? And why exactly do people have such a negative perception of the technology that continues to make our lives so much easier? The USA has the largest interest in AI with over 12 million posts over the past year!

Going by the pains and frustrations of economic headwinds, most people tend to dismiss it as another wasted year. In this milieu, we tend to overlook the tremendous significance of 2018-19 as an inflection point in workforce composition. Those born at the dawn of the new millennium have turned 18 and are at the threshold of the job market. Their share in the workforce is bound to increase over the next decade. The ability of the cohorts to absorb technology changes and the associated cultural implications is getting sharper with each succeeding generation.

Companies around the world are expecting to apply artificial intelligence (AI) within their organisations in the next few years but are lagging in discussions of the ethics around it, new research has found. More than half of the employers questioned in a multi-country opinion survey say their companies do not currently have a written policy on the ethical use of AI or bots, although 21% expressed a definite concern regarding their companies and a potential for the unethical use of AI. Nearly two-thirds (64%) of the employers surveyed expect their companies to be using AI or advanced automation by 2022 to support efficiency in operations, staffing, budgeting or performance, although only 25% are using it now. Yet in spite of this growing trend, 54% of employers questioned say they are not troubled that AI could be used unethically by their companies as a whole or by individual employees (52%). Employees appear more relaxed than their bosses, with only 17% expressing concern about their companies.

A growing number of fintech services are specifically targeting retirees and seniors. From financial management to estate planning, here are the trends driving the nascent space forward. Fintech is often associated with digital tools targeted at tech-savvy millennials. But there is a growing market of fintech companies serving the unique financial needs of Baby Boomers and older retirees -- and for good reason. Q1'19 VC-backed fintech funding dropped but deals remained strong.

Last week, Adam Draper, Managing Director at Boost VC, the Sci-Fi pre-seed fund, tweeted he was looking for Ocean-related tech startups. "The Ocean is the oldest frontier, and it's time to solve global problems with the latest technology," he wrote, followed by "I'm obsessed, and want to profitably help the planet." You might not have pegged Boost VC as an impact investing fund, but that is exactly what they are and have always been, albeit with an emerging tech twist. The term "impact investing" was coined by The Rockefeller Foundation in 2007, which for perspective, is the same year the first generation iPhone was released. It refers to a double or triple bottom line approach in which a company pursues not only a wholesome return in profit but also a positive impact for people and/or the planet.

Millennials are the generation most likely to violate Netflix's terms of service by logging in to use someone else's account, according to a new survey. More than 18 percent of that generation uses someone else's login to stream Netflix, compared to Generation X (9 percent) and Baby Boomers (11 percent). The data comes as Netflix prepares to crackdown on illegal account sharing via new artificial intelligence software, which will be able to analyze which users are logged in and then flag shared accounts. This chart illustrates what proportion of Millennials, Baby Boomers and Generation X stream each of the top three major video streaming services through another person's account The move is expected to recoup major money for the video streaming giant: a separate report from Parks Associates found that by 2021, credentials sharing will account for $9.9 billion of losses in pay-TV revenues and $1.2 billion of over-the-top (OTT) revenues. The survey also found that while 68 percent of people still use traditional cable, nearly 92 percent use video streaming for television, movies, sports or music.