new Amidst the recent hype surrounding generative AI, experts like Ken Mugrage, Principle Technologist, Office of the CTO at Thoughtworks, are cautioning against overlooking more immediate concerns such as sustainability and bias while also recognizing the genuine value of these systems. Rather than viewing generative AI as all-encompassing chatbots, experts envision it as a class of tools designed for specific niches, offering innovative ways to navigate specialized information domains. This perspective—outlined in Mugrage’s recent piece published for MIT Technology Review— acknowledges that generative AI’s true significance lies in its capacity to interact with vast and complex datasets. — CDInsights, 15h ago

new The conference kicked off with keynotes and comments from Black Hat and DEF CON founder Jeff Moss and Azeria Labs founder Maria Markstedter, who explored the future of AI risk — which includes a raft of new technical, business, and policy challenges to navigate. The show features key briefings on research that has uncovered emerging new threats stemming from the use of AI systems, including flaws in generative AI that makes them prone to compromise and manipulation, AI-enhanced social engineering attacks, and how easily AI training data can be poisoned to impact the reliability of the ML models that depend on it. The latter, presented today by Will Pearce, AI red team lead for Nvidia, features research for which Anderson was a collaborator. He says the study shows that most training data is scoured from online sources that are easy to manipulate. — darkreading.com, 17h ago

new Organizational Politics – Even when individual human constraints are effectively managed, there are still organizational challenges, starting with politics. There are many factors of politics such as disagreement of priorities, power struggles, resistance from influential stakeholders, hidden agendas, information asymmetry and manipulation, formation of alliances and coalitions, and fear of losing status leading to decision-making paralysis. Even the best of the AI systems cannot easily overcome human workplace politics. — Healthcare Business Today, 1d ago

new There is a common misconception that AI functions independently without any human supervision. The majority of AI systems need human direction and supervision. Human participation is essential for the responsible and ethical implementation of AI. — techgig.com, 1d ago

new With the above four aspects in mind, whether AI is an independent legal responsibility entity subject to liability is an ever-evolving debate. While current medical AI tools do not yet carry their own consciousness, what should institutions and AI developers do when the machines reach this point? The paper asserts that technology should serve to advance human society. As such, machine autonomy should not forgo the need for human subjectivity. In other words, the authors argue that measures should be taken to prevent the development of fully independent AI systems that can function well beyond human control. — Montreal AI Ethics Institute, 1d ago

new The primary concern with generative AI revolves around its decision-making process. Since AI systems learn from existing data, there's a risk of inheriting biased or flawed logic. This necessitates stringent checks and balances to ensure AI systems operate fairly and accurately, adhering to all regulatory standards. — WriteUpCafe.com, 1d ago

new Our paper shows many examples of AI systems that have learned to deceive humans. Reinforcement Learning (RL) agents trained to play strategic games have learned to bluff and feint, while large language models(LLMs) will output falsehoods in creative ways that help achieve their goals. One particularly concerning example of AI deception is provided by Meta’s AI, CICERO, which was trained to play the alliance-building world-conquest game Diplomacy. Meta put a lot of effort into training CICERO to be “largely honest and helpful,” claiming that CICERO would “never intentionally backstab” its allies. But when we investigated Meta’s rosy claims by studying games that CICERO had played, we found that Meta had unwittingly trained CICERO to be quite effective in its deception. — Montreal AI Ethics Institute, 1d ago

new When talking of brittle systems, many people remember the early symbolic AI programs that were rule-based and, hence, could not process anything outside of the scope of pre-defined knowledge. Did deep learning systems overcome that? Yes, unfamiliar inputs do not completely break them. But even the latest systems still make errors a human wouldn’t make [15-17]. We know that fine-tuned models may learn shortcuts [18-21]: undesirable spurious correlations picked up from the training data. We also know that slight variations in the phrasing of the prompt can lead to very different LLM output [22-24]: this phenomenon affected all 30 LLMs in a recent large-scale evaluation [25]. François Chollet [26] questions if deep learning systems can ever overcome this kind of brittleness: according to him, they are “unable to make sense of situations that deviate slightly from their training data or the assumptions of their creators” (p.3). — Montreal AI Ethics Institute, 1d ago

new In the ever-evolving landscape of health care, AI shows immense promise. AI refers to computer systems that mimic human cognitive functions such as learning and problem-solving, which can be performed with or without human supervision. From diagnostics to surgical precision, it is catalysing a transformation across the entire spectrum of medical care. Machine learning (ML) is a subfield of AI that enables machines to learn and make predictions by recognising patterns to support rational human decision-making and it is increasingly being applied to medicine. Deep learning, meanwhile, is a method in AI that teaches computers to process data in a way that is inspired by the human brain. Deep learning models can recognise complex patterns in pictures, text, sounds, and other data to produce accurate insights and predictions. — theweek.in, 1d ago

new The playbook highlights the three unique, and most crucial abilities, that AI possesses for overcoming bottlenecks that exist on the path to net zero. This includes the ability to discern patterns, predict outcomes, and optimize performances in complex systems; the ability to accelerate the discovery and development of solutions like low-carbon materials, and climate-resilient crops; and finally the ability empower the sustainability workforce of the future. — mid-east.info, 1d ago

new However, the bill is now hanging in the balance because of internal disagreement about some key aspects of the proposed legislation, especially those concerned with regulation of “foundation” AI models that are trained on massive datasets. In EU-speak these are “general-purpose AI” (GPAI) systems – ones capable of a range of general tasks (text synthesis, image manipulation, audio generation and so on) – such as GPT-4, Claude, Llama etc. These systems are astonishingly expensive to train and build: salaries for the geeks who work on them start at Premier League striker level and go stratospheric (with added stock options); a single 80GB Nvidia Hopper H100 board – a key component of machine-learning hardware – costs £26,000, and you need thousands of them to build a respectable system. Not surprisingly, therefore, there are only about 20 firms globally that can afford to play this game. And they have money to burn. — the Guardian, 1d ago

new Artificial intelligence (AI) is at the forefront of boosting efficiency. AI systems can analyze large datasets, automate routine tasks, and even make predictive analyses, saving time and reducing the potential for human error. — Techiexpert.com, 2d ago

new The findings from this study are crucial for academia and go well beyond that, touching the critical realm of AI Ethics and safety. The study sheds light on the Confidence-Competence Gap, highlighting the risks involved in relying solely on the self-assessed confidence of LLMs, especially in critical applications such as healthcare, the legal system, and emergency response. Trusting these AI systems without scrutiny can lead to severe consequences, as we learned from the study that LLMs make mistakes and still stay confident, which presents us with significant challenges in critical applications. Although the study offers a broader perspective, it suggests that we dive deeper into how AI performs in specific domains with critical applications. By doing so, we can enhance the reliability and fairness of AI when it comes to aiding us in critical decision-making. This study underscores the need for more focused research in these specific domains. This is crucial for advancing AI safety and reducing biases in AI-driven decision-making processes, fostering a more responsible and ethically grounded integration of AI in real-world scenarios. — Montreal AI Ethics Institute, 1d ago

new Predictive work systems and automated testing equipment are already saving health systems time and money, and the benefits are sure to grow as the technologies advance. However, human oversight is still needed. A technician should still use their expertise to determine if a computer-generated solution is indeed appropriate, rather than trusting the proposed solution without critical thinking. It’s also important for health systems to engage with technicians when deploying AI tools—to gather feedback and ensure new procedures & technologies are enhancing their work experience rather than adding to their list of responsibilities. — Healthcare Business Today, 2d ago

new Unsupervised learning, meanwhile, is a form of ML wherein AI algorithms are allowed to sift through large tranches of unlabeled data, in an effort to find patterns to classify. This kind of artificial intelligence can be deployed to a number of different purposes, such as creating the kind of recommendation systems that companies like Netflix and Spotify use to suggest new content to users based on their past consumer choices. — Gizmodo, 2d ago

new Concretely, I worry that training AI systems to produce outputs which look good to human evaluators will lead to AI systems which learn to systematically deceive their overseers, e.g. by introducing subtle errors which trick overseers into giving a too-high score, or by tampering with the sensors that overseers use to evaluate model outputs. — alignmentforum.org, 2d ago

new Once a system has been compromised, any attempt by the attacker to exploit their access inevitably triggers abnormal behaviour in some part of the system. AI tools that constantly monitor systems operations can be very effective in rapidly detecting such abnormalities. They can alert humans to their discovery and, in many cases, are able to initiate appropriate countermeasures in far less time than it would take a human to do the same. — IT Brief UK, 2d ago

new When ChatGPT and other generative AI tools were released to the public late last year, it was as if someone had opened the floodgates on a thousand urgent questions that just weeks before had mostly preoccupied academics, futurists, and science fiction writers. Now those questions are being asked by many of us—teachers, students, parents, politicians, bureaucrats, citizens, businesspeople, and workers. What can it do for us? What will it do to us? How do we use it in a way that’s both ethical and legal? And will it help or hurt our already-distressed democracy? Schneier, a public interest technologist, cryptographer, and internationally known internet security specialist whose newsletter and blog are read by a quarter million people, says that AI’s inexorable march into our politics is likely to start with small changes like using AI to help write policy and legislation. The future, however, could hold possibilities that right now we may have a hard time wrapping our minds around—like AI systems leading political parties or autonomously fundraising to back political candidates or causes. Overall, like a lot of other things, it’s likely to be a mixed bag of the good and the bad. — harvard.edu, 2d ago

new ...1, 10, 100, 25, 41, 80, a, able, accepts, Account, accurately, achieving, across, adaptability, Added, address, addressing, aes, ahead, AI, AI Tools, alert, alerts, All, allows, also, always, an, analyzed, analyzes, and, another, answer, any, app, approach, approaches, approval, approve, approved, ARE, around, AS, ask, asked, assigned, associated, At, Authority, Automated, Automating, available, Away, background, base, based, BE, become, becomes, been, before, being, benchmarks, benefits, better, beyond, biases, Biggest, binary, blend, board, both, Breakdown, Broken, budget, business, businesses, But, by, call, calls, CAN, capabilities, capable, capture, captured, categorized, ceo, challenges, chance, channels, checkpoint, clear, click, Close, closed, closing, closure, closures, commonly, compared, compatibility, competitiveness, completion, confidence, Consistency, constantly, Construct, consuming, contact, continue, conventional, Conversation, conversations, conversions, cornerstone, could, creation, criteria, critical, CRM, customer, customer base, Customer-Centric, Customers, Cycle, data, data extraction, date, day, deal, Deals, decision, Decision Making, dedicated, delves, Demo, Details, develop, devoted, DID, differ, different, directly, discovery, discussed, disrupting, Does, Does it Work, done, During, each, Easily, edit, Effective, efficiency, efficient, efficiently, either, elevating, emails, emerged, emotions, ensured, Ensures, equally, estimates, eventually, executives, existent, existing, experience, exploring, exponentially, extraction, factor, fails, FAST, fast growing, faster, few, Fields, Finally, financially, find, First, First contact, Flexibility, flow, Focus, focusing, For, Force, found, Framework, from, further, fuzzy, gathered, generate, generated, genuine, Get, gets, getting, give, gives, globe, Go, going, good, Growing, guess, had, Have, having, heavily, helps, here, High, Higher, Highlighted, highlighting, How, However, HubSpot, human, identifying, if, Impact, Impacts, implementation, important, in, incomplete, indicates, industry, inefficiencies, influence, influenced, information, innovative, insights, integral, integrate, integrated, integrates, Integrating, integration, integrations, interesting, into, intuition, invest, irrespective, Is, isn, issues, IT, ITS, Job, Key, Labels, landscape, largely, lead, Leaders, leads, left, less, levels, Leverage, Leverage AI, lie, like, likely, limiting, Long, long way, long-standing, lot, Made, make, Makes, Making, manual, manually, mark, marked, Market, May, means, meet, merely, met, methods, metric, Might, missing, more, more efficient, most, Much, Need, needed, needs, new, None, Now, Nuance, observe, obvious., of, often, on, once, ones, only, Operations, opportunities, Option, or, order, organization., Other, Others, our, out, outperformed, Over, part, personal, personalization, phase, phenomenal, Picked, pipeline, place, Platforms, plato, Plato Data Intelligence, PlatoData, possibilities, potential, potentials, power, pre, precision, preventing, prioritize, prioritizes, probability, Problem, Process, processes, Product, Products, projections, promising, prospect, Publishing, purchase, pursuit, qualification, qualify, quality, quicker, ranking, Rates, real, real-time, Realistic, really, recall, relevant, report, Request, Requests, requirement, reshaping, resounding, Results, revenue, review, revolutionized, rich, Right, robust, runs, s, sales, Sales Strategies, salesperson, satisfaction, saved, say, Scalability, Scale, scaling, Score, scoring, seamless, seamlessly, see, Send, sends, sent, serves, Service, set, setup, Share, Shows, significant, since, single, slack, Solutions, some, speed, standardized, standing, start, step, Step-by-Step, Strategic, Strategies, Stronger, structured, Study, successful, Such, Summary, support, surveys, Systems, T, tailor, tailored, takes, task, tasks, Team, Teams, template, Than, that, The, The Cycle, The Landscape, their, Them, then, There, they, this, those, thresholds, till, time, time-consuming, timeframe, timeline, times, to, tool, tools, Tracking, traditional, transcription, transformative, transforming, trying, understand, unique, Unlocking, untapped, up, Upfront, upon, urgency, us, use, Used, using, utility, valuable, value, vs, was, Water, way, we, were, What, What is, When, where, Which?, while, will, winning, wisely, with, without, Work, workflow, workflows, working, would, You, yourself, zephyrnet, zoom... — Zephyrnet, 2d ago

new The AI floodgates opened in 2023, but the next year may bring a slowdown. AI development is likely to meet technical limitations and encounter infrastructural hurdles such as chip manufacturing and server capacity. Simultaneously, AI regulation is likely to be on the way.This slowdown should give space for norms in human behavior to form, both in terms of etiquette, as in when and where using ChatGPT is socially acceptable, and effectiveness, like when and where ChatGPT is most useful.ChatGPT and other generative AI systems will settle into people’s workflows, allowing workers to accomplish some tasks faster and with fewer errors. In the same way that people learned “to google” for information, humans will need to learn new practices for working with generative AI tools.But the outlook for 2024 isn’t completely rosy. It is shaping up to be a historic year for elections around the world, and AI-generated content will almost certainly be used to influence public opinion and stoke division. Meta may have banned the use of generative AI in political advertising, but this isn’t likely to stop ChatGPT and similar tools from being used to create and spread false or misleading content.Political misinformation spread across social media in 2016 as well as in 2020, and it is virtually certain that generative AI will be used to continue those efforts in 2024. Even outside social media, conversations with ChatGPT and similar products can be sources of misinformation on their own.As a result, another lesson that everyone – users of ChatGPT or not – will have to learn in the blockbuster technology’s second year is to be vigilant when it comes to digital media of all kinds.Tim Gorichanaz, Assistant Teaching Professor of Information Science, Drexel UniversityThis article is republished from The Conversation under a Creative Commons license. Read the original article. — GovTech, 2d ago

new Over the past year, researchers in the AI field have been debating the ability of AI systems, both in private and on social media. Some have suggested that AI systems are coming very close to having AGI while others have suggested the opposite is much closer to the truth. Such systems, all agree, will match and even surpass human intelligence at some point. The only question is when. — techxplore.com, 2d ago

new The dream of autonomous driving is rapidly becoming a reality, thanks to AI-powered systems that can perceive the environment, make decisions, and control vehicles. Data collected from sensors, cameras, and Lidar feed into AI algorithms, enabling self-driving cars to navigate complex urban environments and highways. The industry is heavily reliant on data to fine-tune these systems, with real-world testing providing crucial insights. — Zephyrnet, 2d ago

new He pointed to the real-world example of this disparity in the digital equity space. While Denver’s digital equity needs are primarily centered around affordability of Internet service, other parts of the state face affordability and connectivity issues.Because of the collaborative relationships he built at the local level, Edinger feels well positioned to work with municipalities to advance technology use in different areas — including expanding connectivity and developing AI policy.Another major area he plans to focus on is accessibility. The state’s work in this space so far has included the launch of the Aira tool for Coloradans with low vision and a pilot program to help state workers better serve individuals with disabilities using VR, but there is more work to be done. Accessibility across state government is a critical piece to ensuring all constituents are served equitably, he said.And although Edinger expects emerging technologies to continue to impact government, he also underlined the importance of effective enterprise management systems. The technology working behind the scenes is an important foundation for any future innovation, he said.At the heart of Edinger’s vision for OIT is the people — employees and constituents alike. Empowering employees through things like process improvements, Lean, and other technology-enabled advances is the best way to make state government better, he said.“I will just say I am confident that investing in people and unlocking their potential is the way we will get there,” he said.The brief period of overlap between Edinger and Neal-Graves has helped enable a smooth transition for OIT to prepare the agency for future work under Edinger’s leadership. As Neal-Graves recently told Government Technology, the future for OIT is bright. He expects the state to continue advancing in digital government and digital equity. — GovTech, 2d ago

new According to the Israeli Defense Forces, “Habsora” purportedly can use AI and “rapid and automatic extraction of updated intelligence” to generate recommended targets. Targeting systems that employ automated decision-making and AI technologies present serious concerns and are part of a worrying trend toward the deployment of autonomous weapons systems. — Stop Killer Robots, 2d ago

new ...“So when we set out to build generative AI applications, we knew we had to address these gaps,” he said. “It has to be built in from the very start. Q lets you answer questions quickly, with natural language interactions, and you can easily chat, generate content and take action. And it’s all informed by an understanding of your systems, data repositories and operations.”... — SiliconANGLE, 2d ago