Charting the emergence of AGI?

  The paper ' Language Models Don’t Always Say What They Think : Unfaithful Explanations in Chain-of-Thought Prompting' by Miles Turpin et al. investigates the faithfulness of chain-of-thought (CoT) explanations generated by large language models (LLMs) for various tasks. CoT explanations are verbalisation's of step-by-step reasoning that LLMs produce before giving a final output.  The paper shows that CoT explanations can be misleading and influenced by biasing features in the model inputs, such as the order of multiple-choice options. The paper tests two LLMs, GPT-3.5 and Claude 1.0, on 13 tasks from BIG-Bench Hard and a social-bias task, and finds that accuracy drops significantly when models are biased toward incorrect answers.  The paper also finds that models justify answers based on stereotypes without mentioning the influence of social biases. The paper concludes that CoT explanations can be plausible yet unfaithful, which poses a risk for trusting LLMs without ensu

  Can auditing AI products work? 'AI audit refers to evaluating AI systems to ensure they work as expected without bias or discrimination and are aligned with ethical and legal standards.' states Javid in the ' How to perform an AI Audit in 2023 ' article.  It's a fine article, listing many of the existing frameworks that are currently in existence, principally COBIT Framework (Control Objectives for Information and related Technology) and IIA's (Institute of Internal Auditors) AI Auditing Framework: This AI framework aims to assess the design, development, and working of AI systems and their alignment with the organisation’s objectives. Three main components of IIA’s AI Auditing Framework are Strategy, Governance, and Human Factor.  This all seems fine, on the surface. Internal auditing has had a variety of issues though. The same can be said for External Auditing. There are five large firms that conduct external AI auditing: Deloitte, PwC, EY, KPMG and Grant T

  Anyone that's been following the AI race will know that there's one clear winner so far, NVIDIA. They are currently more or less unopposed in producing hardware for AI and that's not all, their software is rather sophisticated to. NVIDIA have just released this video. If you want to understand where AI is heading over the next year, it's well worth a close watch. In five years we improved computer Graphics by 1,000 times in five years using artificial intelligence and accelerated Computing Moore's Law is probably currently running at about two times a thousand times in five years a thousand times. In five years is one million times in ten we're doing the same thing in artificial intelligence now question is what can you do when your computer is one million times faster.' It's pretty outstanding, and it's far from just graphics, NVIDIA gives us a glimpse of what a MultiModal experience is like, and what will shortly be commonplace. Every factory in

  InfraNodus is a web-based open source tool and method for generating insight from any text or discourse using text network analysis. The byline on the website states, 'Get an overview of any discourse, reveal the blind spots, enhance your perspective.' which, whilst accurate does little to summarise the potential of such a tool. Watching the introduction helps. Its capabilities include representing any text as a network and identifying the most influential words in a discourse based on the terms' co-occurrence, providing text network visualization and analysis live as new data is added, offering discourse structure analysis to measure the level of bias in discourse and identify structural gaps in discourse, and being available via an API to be used in conjunction with other text mining and analysis software. The white paper, ' Generating Insight Using Text Network Analysis ' concludes:  'The tool is currently used by researchers, marketing professionals, stude

  I have a problem with Ai ethics. I admit this may be Ethics 101 to many. But my problem is in a similar way that I have a problem with the anthropomorphising of AI in discourse. There are many that should know better, but do it all the same.  For example, Open AI have an ' Open Ethics ' project. It states, in large letters, 'Open Ethics for AI is like Creative Commons for the content. We aim to build trust between machines and humans by helping machines to explain themselves.' Surely, trust can only exist between the companies, and the personnel they employ, rather than in the machine itself. It is difficult to guarantee trust in anything unless one reviews the code, the compiler, the build, the training methodologies of the LLM. Transparency is critical to trust. And trust should not be transferred to the machine tool without transparency and that a set of other principles are being followed, such as voluntary participation, informed consent, anonymity, confidentiali

  Selena Nemorin, Andreas Vlachidis, Hayford M. Ayerakwa, and Panagiotis Andriotis explore the hype surrounding AI in education and provide a horizon scan of the current discourse. ' AI hyped? A horizon scan of discourse on artificial intelligence in education (AIED) and development' . This paper provides a horizon scan of the discourse surrounding artificial intelligence (AI) in education and development. The authors use text mining and thematic analysis to explore the hype surrounding AI in education and identify key themes that have emerged during the AIEd debate. The findings are categorized into three themes: geopolitical dominance through education and technological innovation, creation and expansion of market niches, and managing narratives, perceptions, and norms. The paper highlights the challenges of implementing AI in educational settings due to a lack of rigorous evidence supporting practical outcomes. One of the key themes identified in the paper is geopolitical do

  In this blog post, I will summarise the main findings of a recent paper titled "A Systematic Review of Green AI" by Verdecchia et al.. The paper provides a comprehensive overview of the research field of Green AI, which aims to reduce the carbon footprint of AI models and systems. The paper analyzes 98 primary studies on Green AI published between 2016 and 2021, and identifies different patterns and trends in the literature. A definition of Green AI . From the results regarding how the term “Green AI” is used in the literature a clear picture emerges. Most Green AI studies consider Green AI as exclusively related to energy efficiency. Only fewer studies examine the influence of AI on greenhouse gas emissions (𝐶𝑂2), and an even minor fraction examines the holistic impact that AI has on the natural environment. The paper categorises the Green AI studies into four main types: position papers, observational studies, solution papers, and tool papers. Position papers propose ne