Ross McLean’s Post

In a new article "Collaborating” with AI: Taking a System View to Explore the Future of Work" article, Callen Anthony NYU Stern School of Business Beth Bechky University of California, Davis and Anne-Laure Fayard Nova School of Business and Economics published in the September-October issue of Organization Science examine traditional theories of technology in organizations to frame new directions for the study of what it means to work “with” artificial intelligence (AI). Acknowledging the unique nature of AI compared to technologies studied in the past, the research team ultimately concludes that a better understanding of AI, work, and collaboration requires a broader perspective that looks at a comprehensive system rather than solely its users. Why study this now: As AI continues to permeate society, many organizations have struggled to effectively develop and implement new technologies into their existing ecosystems. The authors examine why traditional perspectives of technology as a tool and as a medium might suppress important material, symbolic, and relational dimensions of AI. Instead, they propose conceptualizing AI as an active counterpart in a system of interactions to unpack and understand its role in organizational processes, as well as the unintended consequences that might emerge from all stakeholders working with AI.   Key Takeaways: ● Studying a system vs. just users: By studying a system, or a set of related components that work together to perform functions, scholars can analyze the interactions and relationships between different human actors in occupations and organizations as well as AI algorithms and their outputs. ● Current perspectives on technology, as a tool and as a medium, may struggle to counter the hype about human-AI collaboration because of the way they conceive of technology and how people work with it. ● AI has three specific material characteristics, constant change, invisibility, and inscrutability, that challenge traditional approaches to the study of technology and work and limit applicability of previous findings. ● To effectively provide a more grounded and deeper picture of AI, researchers need rich data collected through relational ethnography.   Key insight: “In a system view, the focus is not on individual actors but on how knowledge creation and work result from interactions between different agents,” noted the authors. “Taking a system lens and acknowledging the role of AI as an actor in that system allows us to attend to broader consequences when it comes to unpacking the power dynamics between producers, implementers, and users of AI and the ethical issues that might arise.”

What is AI’s perspective on the role of intuition in decision-making? ✍️ \[SUMMARY\] AI’s Perspective on the Role of Intuition in Decision-Making 🤖📚🧠💡🔍🤔🎯✅💬 Artificial Intelligence \(AI\), despite its vast capabilities, has its limitations when it comes to understanding the role of intuition in decision-making. While AI excels in analyzing and processing vast amounts of data, it lacks the human ability to rely on intuition, emotions, and \[…\] ✍️ AI's Perspective on the Role of Intuition in Decision-Making 🤖📚🧠💡🔍🤔🎯✅💬 Artificial Intelligence \(AI\), despite its vast capabilities, has its limitations when it comes to understanding the role of intuition in decision-making. While AI excels in analyzing and processing vast amounts of data, it lacks the human ability to rely on intuition, emotions, and gut feelings. 🤖🎯🔍 #AI #Intuition #DecisionMaking #DataAnalysis #HumanPerspective #Technology #DataPoweredAI Intuition holds a significant place in human decision-making, but how does AI perceive it? AI primarily functions based on algorithms and data-driven analysis. Although it can analyze enormous datasets and provide logical conclusions, it struggles to understand the nuances of human intuition. The rigidity of AI's programming may hinder its ability to consider subjective factors influencing decisions. 🔍💡💭💻 AI's programming is focused on following a set of predefined rules and parameters, aiming to minimize errors and optimize outcomes. This approach often sidelines the value and impact of intuition in decision-making. Intuition, on the other hand, involves tapping into unconscious processes and personal experiences, which AI lacks. 📚🤔🎯 Human decision-making, with intuition as a core component, often considers factors like emotions, cultural context, and past experiences. These elements are essential in making complex decisions, especially ones that involve ethical considerations, creativity, and empathy. AI, being devoid of emotions and subjective experiences, relies solely on objective data analysis, potentially limiting its understanding and evaluation of intuition's role. 💡💭🔍 With advancements in AI, researchers aim to develop models that simulate human intuition. However, replicating human-like intuition remains a challenge. While AI can assist in decision-making by providing data-driven insights, incorporating the human touch and intuition is crucial for making well-rounded and comprehensive choices. The symbiotic collaboration of AI and human intuition can lead to more informed and effective decisions. 🤷♀️🙋♂️💬 **Frequently Asked Questions \(FAQs\)** 1. Does AI completely disregard intuition in decision-making? AI doesn't disregard intuition but struggles to comprehend and incorporate it due to its inherent limitations in understanding human emotions and subjective experiences. 2. Can AI learn and evol

How Artificial Intelligence Might be Worsening the Reproducibility Crisis in Science and Technology https://lnkd.in/dYBcJUM2 AI News, AI, AI tools, Innovation, itinai.com, LLM, LucianoSphere (Luciano Abriata, PhD, t.me/itinai, Towards Data Science - Medium **AI Solutions for Middle Managers** **How AI Impacts Scientific Research and Technological Applications** Artificial Intelligence (AI) has become an essential tool in scientific research, but concerns are on the rise about its misuse leading to a reproducibility crisis in science and its technological applications. It's crucial to understand the fundamental issues contributing to this effect and explore specific challenges related to AI-based science. **Reproducibility in Science and AI-specific Issues** Reproducibility is a key principle in the scientific method, requiring that experiment results, including those from AI models, must be fully replicable and repeatable. However, AI-based science faces challenges such as data quality, modeling transparency, and risks of data leakage. **Data Quality and Availability** AI models heavily rely on vast amounts of reliable, well-curated data for training, making data quality and availability crucial for scientific accuracy. **Modeling Details and Transparency** Describing and documenting the methods and models used in a study is essential for reproducibility, but this can be complex for AI models due to their intricacy. **Risks of Data Leakage and Manipulation** Data leakage and manipulation can skew AI systems and lead to inaccurate results, posing a significant challenge in AI model development. **Challenges in Creating AI Models for Real-World Conditions** AI models must accurately reflect real-world conditions to perform effectively when deployed in practical scenarios. **Setting up Standards to Address Challenges and Problems** Researchers have proposed a checklist of standards for reporting AI-based science, emphasizing the need for transparency and rigorous standards in AI-based research. **Practical AI Solutions for Middle Managers** If you aim to leverage AI to stay competitive and enhance your company's operations, consider the following practical solutions: 1. **Identify Automation Opportunities**: Discover key customer interaction points that can benefit from AI. 2. **Define KPIs**: Ensure your AI initiatives have measurable impacts on business outcomes. 3. **Select an AI Solution**: Choose tools that align with your needs and offer customization. 4. **Implement Gradually**: Begin with a pilot, gather data, and expand AI usage judiciously. For AI KPI management advice, connect with us at [hello@itinai.com](mailto:hello@itinai.com). To stay updated on leveraging AI, follow our Telegram channel or Twitter. **Spotlight on a Practical AI Solution** Explore t...

New Post: Artificial and Natural Intelligence: A peek - https://lnkd.in/er6bq4wj News Network Title: AI and Natural Intelligence: Understanding the Difference Introduction:Artificial Intelligence - AI - and natural intelligence represent two distinct forms of intelligence. While natural intelligence refers to the cognitive abilities possessed by humans and animals, AI refers to the simulation of human intelligence in machines. This article aims to explore the differences between AI and natural intelligence, highlighting their unique characteristics and capabilities. Natural Intelligence:Natural intelligence refers to the cognitive abilities exhibited by living organisms, particularly humans. It encompasses a wide range of capabilities, including perception, learning, reasoning, problem-solving, and decision-making. Natural intelligence is characterized by its adaptability, creativity, and consciousness - Hassabis et al., 2017 -. It is deeply rooted in the complex structure and functioning of the human brain. Artificial Intelligence:AI, on the other hand, refers to the development of machines and computer systems that can perform tasks that typically require human intelligence. AI systems are designed to mimic human cognitive processes, such as learning, reasoning, and problem-solving, through algorithms and computational models. AI can be further classified into narrow AI, which is designed for specific tasks, and general AI, which possesses human-like intelligence across a wide range of domains - Russell et al., 2021 -. Differences between AI and Natural Intelligence:a. Origin: Natural intelligence arises from the complex biological structure and functioning of the human brain, while AI is created and developed by humans through programming and algorithms.b. Learning: Natural intelligence is characterized by the ability to learn from experiences, adapt to new situations, and acquire knowledge over time. AI systems learn from data through training algorithms and can improve their performance with more data and iterations.c. Creativity: Natural intelligence exhibits creativity, imagination, and the ability to generate novel ideas and solutions. While AI can generate outputs based on existing patterns and data, it lacks the inherent creativity and imagination of natural intelligence.d. Consciousness: Natural intelligence is associated with consciousness, self-awareness, and subjective experiences. AI, as of now, does not possess consciousness or subjective experiences and operates purely based on algorithms and data processing.e. Limitations: Natural intelligence is highly adaptable and capable of handling a wide range of tasks, even in unfamiliar situations. AI, however, is limited to the specific tasks it is designed for and may struggle with tasks outside its programmed capabilities. Synergies between AI and Natural Intelligence:While AI and natural intelligence have distinct ch

Harvard-BCG Study (Working Paper, Sep. 2023): Does AI Boost Knowledge Worker Productivity & Quality? Depends. A. If Within AI's Capabilities, Yes "For each one of a set of 18 realistic consulting tasks within the frontier of AI capabilities, consultants using AI were significantly more productive (they completed 12.2% more tasks on average, and completed tasks 25.1% more quickly), and produced significantly higher quality results (more than 40% higher quality compared to a control group)." B. If Outside, No "For a task selected to be outside the frontier, however, consultants using AI were 19 percentage points less likely to produce correct solutions compared to those without AI." "Outside of it, humans relied too much on the AI and were more likely to make mistakes." C. Lower-Ranked Performers Benefited More from AI "By segmenting subjects exposed to one of the two AI conditions into two distinct categories — top-half-skill performers (those ranking in the top 50% on the assessment task) and bottom-half-skill performers (those in the bottom 50%) — we observed performance enhancements in the experimental task for both groups when leveraging GPT-4. When comparing the two groups, though, we see the bottomhalf-skill performers exhibited the most substantial surge in performance, 43%, compared to the top-half-skill subjects, 17%." https://lnkd.in/gsnGv6GY

Very interesting research on the effects of AI in day to day work (mainly office work). Key findings: ➡️ The study involved 758 consultants from Boston Consulting Group, who were randomly assigned to experimental conditions to test the impact of GPT-4 AI on their performance on realistic knowledge work tasks. For tasks designed to be within the capabilities of GPT-4 ("inside the frontier"): ➡️ Consultants using GPT-4 were significantly more productive - they completed 12.2% more tasks and completed tasks 25.1% more quickly compared to a control group without AI access. ➡️ Consultants using GPT-4 produced responses that were rated over 40% higher quality by human evaluators compared to the control group. ➡️ Consultants across all skill levels benefited from using GPT-4, but the lower-skilled consultants had the largest performance increases of 43% compared to 17% for the higher-skilled consultants. For a task designed to be outside GPT-4's capabilities ("outside the frontier"): ➡️ Consultants using GPT-4 were 19 percentage points less likely to produce correct solutions compared to those without AI. ➡️ Even when consultants using GPT-4 produced incorrect solutions, the quality of their recommendations was rated higher by human evaluators. In summary, the study provides experimental evidence that GPT-4 can significantly enhance knowledge workers productivity and quality for tasks within its capabilities, while degrading performance on tasks outside its capabilities. It highlights the "jagged frontier" of AI capabilities that humans must navigate.

🕵️♂️The hidden flaw in Generative AI…Can your model be trusted? Your model is only as good as your training data. If we don’t get that part right, we’ll be prone to different types of biases: 📊Algorithm  🗣️Linguistic  🗞️Confirmation And many more… So, why does this matter? When your data is inaccurate, you risk unreliable outputs. Imagine deploying a language model to assist in hiring. If it’s trained on biased data, it might favor certain demographics over others. Not ideal, right? Bias can impact business decisions, product development, and even customer relationships. It’s not just a technical issue, it’s a business risk. Let’s focus on improving data quality. Better data means more reliable AI. And reliable AI can lead to groundbreaking innovations. Stay ahead. Invest in your data. 🚀 Read more in our latest blog post: https://lnkd.in/dmms82xt #GenerativeAI #MachineLearning #Data #Innovation

Medium 6 New Studies Put AI to the Test MIT scholars offer data-based evidence to guide employers, developers and policymakers GenAI search results are distrusted. Job descriptions generated by large language models are efficient, but don’t always lead to more hires. And adding “beneficial friction” can help AI users to be more thoughtful and accurate. These are some of the latest findings of research projects under way at the MIT Initiative on the Digital Economy (IDE). At the 2024 Annual Conference, an online event presented for members during the week of May 20, many speakers described ongoing AI studies they hope will provide practical guidance to those building and using GenAI.

An interesting paper measuring the efficiency of leverage LLM. TL;DR - Know the tasks then AI can be a great tool to improve efficiency. - If the task was outside of your competency, then using LLM as a tool may harm your performance. (Do your homework first?) Corollary: - It would seem that the more specific you can get with your question/ask to LLM (as you know in more detail about how to get to result), the better the word prediction becomes. - Almost can picture this as a knowledge graph where the closer you are to the knowledge, the less drift you may have traversing the graph. Experiment setup: - The study in question was conducted with 758 BCG consultants scoring over 18 consulting tasks. Finding: - Consultants using AI were significantly more productive (they completed 12.2% more tasks on average, and completed task 25.1% more quickly), and produced significantly higher quality results (more than 40% higher quality compared to a control group) - Those below the average performance threshold increased by 43% and those above increased by 17% compared to their own scores - For a task selected to be outside the frontier, however, consultants using AI were 19 percentage points less likely to produce correct solutions compared to those without AI Source: https://lnkd.in/gZJCWTn2 #llm #ai #study