Pravir Malik, Ph.D.

The Bill & Melinda Gates Foundation has several assessment of the current K-12 instructional material--particularly powered by tech, market, which I cannot agree with them more. To highlight two that resonate with me the most: 1. We do not have sufficient evidence on what product features -- particularly powered by AI-- and enactment practices work to foster student motivation, engagement, and persistence (MEP), particularly for Black, Latino, and all students experiencing poverty. 2. The current market does not adequately highlight what works when it comes to effectively motivating and engaging students, and does not reward products that shows evidence of increased student learning. This failure means product developers have little guidance on and motivation to build and sustain products that foster student MEP while leaving buyers unaware of how to pick higher quality instructional materials and tech products. Appreciate the K-12 team identifies these critical issues and work to address them! Andrea Cobb ; Kysie Miao Jensen ; Jessica Korf Beaver; Bryan Richardson

31

2 Comments

I’m honored to be featured in Higher Education Digest, discussing how AI can be a force for positive social impact and equitable growth. 🚀 In the interview, we dive into empowering the next generation of AI entrepreneurs and the importance of operationalizing AI and building tech that benefits everyone, not just a select few. Thank you to the team at Higher Education for the platform to share my thoughts on this critical topic and to my colleagues at #CMU for providing me a great academic environment to teach and learn at the same time. Ramayya Krishnan, Andy Wasser, Jackie Speedy, Shryansh Mehta Check out the full interview below. #AI #SocialImpact #Innovation #Entrepreneurship #TechForGood

122

3 Comments

Leveraging AI to meet emotional needs, insights from recent research A recent study published in the *Proceedings of the National Academy of Sciences* has unveiled fascinating insights into how AI can effectively enhance emotional support, even outperforming untrained humans in making recipients feel heard. Conducted by researchers from the USC Marshall School of Business, the study not only underscores AI's potential in understanding and communication but also highlights the perceptual biases that diminish its perceived value. Despite AI's adeptness at detecting emotions and offering hope, the study reveals a significant bias: individuals feel less heard when they know a message is AI-generated. This "response bias" suggests that the effectiveness of AI in emotional support roles might be compromised by its identification as non-human. The research also touches upon the "uncanny valley" effect, where the awareness of AI's role in empathetic responses can trigger discomfort. For CIOs, CTOs, and CAIOs, these findings point to a crucial aspect of digital transformation—how AI is integrated and presented in organizational structures can significantly influence its acceptance and effectiveness. As AI becomes a more integral part of our social and professional ecosystems, understanding and mitigating biases against AI could enhance its utility, making it a valuable tool for fostering human connection in increasingly digital landscapes. Action Plan — Educate and Train Increase awareness and training sessions within your teams to reduce biases against AI-generated support. Strategic Deployment Integrate AI in ways that complement human roles, emphasizing its strengths without overshadowing the human element. Monitor and Adapt Regularly assess user feedback to adapt AI tools for better personalization and effectiveness. #DigitalTransformation #AI #EmotionalIntelligence #Leadership #TechForGood https://lnkd.in/gekVckKy

4

2 Comments

Now AI is real and deeper insights of what the labour market impact of AI could be are emerging. Here is one research (paywall). The abstract says it all - 46% of the jobs!

2

Generative AI is evolving at record speed while CEOs are still learning the technology’s business value and risks. Here, McKinsey and Company offers some of the generative AI essentials.

11

Terence Tao, a mathematics professor at UCLA, is a real-life superintelligence The “Mozart of Math,” as he is sometimes called, is widely considered the world’s greatest living mathematician. He has won numerous awards, including the equivalent of a The Nobel Prize for mathematics, for his advances and proofs. Right now, AI is nowhere close to his level But technology companies are trying to get it there Recent, attention-grabbing generations of #AI—even the almighty #ChatGPT—were not built to handle mathematical reasoning They were instead focused on language: When you asked such a program to answer a basic question, it did not understand and execute an equation or formulate a proof, but instead presented an answer based on which words were likely to appear in sequence For instance, the original ChatGPT can’t add or multiply, but has seen enough examples of algebra to solve x + 2 = 4: “To solve the equation x + 2 = 4, subtract 2 from both sides” Now, however, OpenAI is explicitly marketing a new line of “reasoning models,” known collectively as the o1 series, for their ability to problem-solve “much like a person” and work through complex mathematical and scientific tasks and queries

13

2 Comments

"We already know that possibly tens of thousands of students and educators are using AI both in and out of the classroom," State Superintendent Chris Reykdal said. "We now get to put some shape and definition around this usage by embracing it with a human-centered approach." Appreciate seeing how states and districts are tackling challenges & opportunities in this new area of #AIinEducation, together. The team at the National Center on Education and the Economy has put together an actionable and accessible framework to help others think about the near future and long-term in their context. https://bit.ly/4cM2rjj #TeachAI #FutureOfEd

8

It was great meeting up with Abdullah Almaatouq at Sloan to learn about his incredible research, and I also enjoyed listening to a book for my #AcademicRunPlaylist! Today I listened to "Free Enterprise" by Lawrence Glickman. While many of us associate the concept of "free enterprise" with the neoliberal turn of 1980, or maybe as the reaction against the New Deal decades earlier, Glickman shows how the roots of the term are much deeper. Starting in the 1800s as an evolution of the "free labor" concept, which contrasted with slave labor, the amorphous concept of free enterprise was, at its heart, an attempt to re-entrench power for those that already had it. As this book shows, attempts to pin down a definition all failed. For conservatives it has always meant a combination of laissez-faire when applied to business and a complete repeal of any attempts to legislate nondiscrimination or prescribe taxes of any kind. The book returns to this point again and again, which is one reason it could have been much shorter. Beyond that, there is some off topic (and incorrect) criticism of supply chain complexity which distracts from the overall message. Still, if you're interested in where this concept came from and the history of the neoliberal movement, this is a good place to start. https://lnkd.in/eRPCPfEN

18

🧠💡 Exciting new research from UC San Francisco reveals that young adults with higher levels of inflammation—often linked with factors like obesity, inactivity, and stress—may face reduced cognitive abilities by midlife. This groundbreaking study, published in Neurology on July 3, suggests that habits formed in early adulthood could significantly impact cognitive health and the risk of dementia later in life. A proactive approach to reducing inflammation through lifestyle changes could be a key strategy in preventing cognitive decline. #Neurology #CognitiveHealth #Inflammation #UCSF Read more: https://lnkd.in/eshQexd8

3

"Ott's Product Management class at CMU-SV is structured to give students a holistic understanding of the product lifecycle. The course emphasizes a balanced approach, ensuring that students not only grasp the theoretical aspects of product management but also gain practical experience. By working on real projects, students are taken through each phase of the product lifecycle, fostering a deep and comprehensive understanding." This excellent article by Utkarsh Khandelwal and Philip Geist gives you a peek at what it is like to take the Product Management course by Adrian C. O. at Carnegie Mellon University - Integrated Innovation Institute. https://lnkd.in/gHDpc8gR

18

Our society has an obsession with quantifying everything, often applying measurements and numerical values where they aren’t necessary. Mary Helen Immordino-Yang, a neuroscientist and professor at USC, argues that we’ve taken this approach with intelligence, and it might not be the right path. Immordino-Yang uses standardized testing as an example, explaining how modern-day education systems equate high test scores with high intelligence levels. However, these tests usually only measure a student’s ability to perform under specific conditions, focusing narrowly on memorization and regurgitation of predetermined answers. Instead, Mary Helen suggests a more dynamic concept of intelligence, considering a child’s ability to navigate complex situations, understand new information, and innovate in real-time. This kind of intelligence is adaptive and essential for societal progress, as it allows for a more well-rounded perception of the world and situations that occur within it.

16

This looks promising. I hope that some of the focus will be spent on confronting the significant risks that generative AI poses for communities and specific vulnerable populations via the ability for bad actors to more easily produce health disinformation and infodemics during epidemics and other public health emergencies. As I presented during a talk in Fall 2023 on its implications for vaccination uptake, with generative AI, the tactics of such disinformation efforts likely will not change much, but the technology enables the volume of such efforts to increase.

4

4 Comments

From Data to Drugs: How CADD is Shaping the Future of Medicine The drug discovery landscape is evolving rapidly, driven by technological advancements like Artificial Intelligence and Machine Learning. These tools are not just buzzwords; they’re revolutionizing how we discover new drug candidates. 💡 Picture this: Traditional drug discovery takes over a decade and costs billions. What if we could slash that time and cost without sacrificing quality? That is what Computer-Aided Drug Discovery (CADD) does—it accelerates early-stage discovery. CADD can screen thousands of compounds against targets in days, not months, helping us focus on the most promising candidates right from the start. It goes beyond screening—CADD fine-tunes drug candidates, optimizing their binding affinities and predicting side effects before we even step into the lab. While CADD isn’t a replacement for experimental synthesis and biological screening, it makes these processes more efficient, streamlining the journey from data to drugs. The future of drug discovery is already here, and it's computational. 🌐 I’m hosting a one-week virtual training on Molecular Docking—perfect for beginners! Say goodbye to outsourcing and start performing your own analysis. Limited spots are available, register now: https://lnkd.in/duHbeXta #TheDockingDoctor #DrugDiscovery #Pharma #MolecularDocking

15

It has been almost five years since Prashant Natarajan reached out to me on LinkedIn about collaborating on the book Demystifying AI for the Enterprise and more than two years since this interview article by Carly Ochs for the Carnegie Mellon University - Integrated Innovation Institute blog. This excellent question and answer still holds in the age of Generative AI: "Q: It seems there is still a considerable amount of apprehension or, perhaps, confusion around AI. In your opinion, what is the biggest misconception or myth about AI? A: There are many myths. The biggest one, in my opinion, is that it is a panacea. Just like any other technology, it is all about how you make use of it. It has many benefits, but it is not needed in many instances. Good statistical tools probably suffice in many cases.  It is essential to analyze the problem to ensure you need AI and have the quality data required to use it. I have seen several projects where the management dictates the use of AI without understanding the availability of data to make the impact or underestimates the effort to get clean, usable data." #genai #generativeai https://lnkd.in/g9Sun_k6

18

2 Comments

In 2005, I came to Purdue University to validate a model I had developed to address complex challenges in open networks...situations in which no one can tell anyone else what to do. We can't address today's complex challenges by following Industrial Age approaches. We can't optimize a complex adaptive system or perform root-cause analysis on a wicked problem. We cannot "decide and delegate" our way to a better world. ----------- "The greatest danger in times of turbulence is not the turbulence; it is to act with yesterday’s logic." Peter Drucker ----------- We need new approaches to collective action and experimentation. We need to develop new skills in thinking, behaving, and doing work together. As my colleague Doug Smith reminds us, we must learn new behaviors and skills to "design and do." Here is a map to share my connections with Strategic Doing practitioners in the past few weeks. Our small team at Purdue University incubated this discipline for generating replicable, scalable, and sustainable solutions to complex challenges. We ran testbeds to address complex challenges, from developing new collaborations among NASA's life scientists to reducing teenage homicides in Flint, Michigan. We developed a clear, concise protocol for collective action. Multiple streams of scholarly research -- from cognitive psychology to strategic management and behavioral economics -- explain why it works. We are spreading this discipline to practitioners worldwide as fast as possible. Join us.

45

5 Comments

How do organizations justify AI Ethics initiatives? They can emphasize either loss aversion or value generation. In our work with at the Notre Dame-IBM Tech Ethics Lab we studied 30 organizations to understand how they think about ROI for AI ethics. We found two general approaches. Sometimes we found a "loss aversion" approach, where organizations justified investing in ethical AI in terms of savings from avoiding regulatory fines or avoiding losing customers. Loss aversion approaches are reactive to external entities like regulators and competitors, but they are more certain, in the sense that they are more easily quantifiable and measurable. The alternative is a "value generation" approach. When justifying AI ethics through value generation, organizations emphasized building capabilities, competitive advantage, and a sincere, internalized commitment to ethical AI. The value generation approach is more proactive, but requires a long-term outlook and organizations need to be more comfortable with uncertainty and intangible ROI. Read more about the study here: https://lnkd.in/gGvBJRif Heather Domin, PhD Brian C. Goehring Francesca Rossi Marialena Bevilacqua Marianna B. Ganapini

125

7 Comments

Loving some of the examples and images Omar Khawaja uses to describe and clarify the very confused space of AI risk together with the other panelists Arun Pamulapati and Jacqueline Lebo of the FAIR Institute workgroup on AI risk . If you missed the live session, check out the replay for true insights on AI risk on FAIRinstitute.org

41

4 Comments

Generative AI is truly remarkable. Yet, it is often misunderstood as having capabilities far beyond its actual scope. The apparent intelligence of genAI models is a reflection of sophisticated pattern recognition rather than true cognitive abilities. In this interview write-up Rodney Brooks (co-founder of iRobot) emphasizes the importance of recognizing these limitations to avoid unrealistic expectations and misplaced trust in AI technologies. For example, there's no point building an LLM for warehouse robots. One can do that much more efficiently by connecting the robot to data coming from the warehouse software via existing mathematical models. https://lnkd.in/e6R8ND4W

2