Data Skeptic

Our guest today is Vid Kocijan, a Machine Learning Engineer at Kumo AI. Vid has a Ph.D. in Computer Science at the University of Oxford. His research focused on common sense reasoning, pre-training in LLMs, pretraining in knowledge-based completion, and how these pre-trainings impact societal bias. He joins us to discuss how he built a BERT model that solved the Winograd Schema Challenge.

Today, We are joined by Petter Törnberg, an Assistant Professor in Computational Social Science at the University of Amsterdam and a Senior Researcher at the University of Neuchatel. His research is centered on the intersection of computational methods and their applications in social sciences. He joins us to discuss findings from his research papers, ChatGPT-4 Outperforms Experts and Crowd Workers in Annotating Political Twitter Messages with Zero-Shot Learning, and How to use LLMs for Text Analysis.

In this episode, we are joined by Carlos Hernández Oliván, a Ph.D. student at the University of Zaragoza. Carlos's interest focuses on building new models for symbolic music generation. Carlos shared his thoughts on whether these models are genuinely creative. He revealed situations where AI-generated music can pass the Turing test. He also shared some essential considerations when constructing models for music composition.

Hongyi Wang, a Senior Researcher at the Machine Learning Department at Carnegie Mellon University, joins us. His research is in the intersection of systems and machine learning. He discussed his research paper, Cuttlefish: Low-Rank Model Training without All the Tuning, on today's show. Hogyi started by sharing his thoughts on whether developers need to learn how to fine-tune models. He then spoke about the need to optimize the training of ML models, especially as these models grow bigger. He discussed how data centers have the hardware to train these large models but not the community. He then spoke about the Low-Rank Adaptation (LoRa) technique and where it is used. Hongyi discussed the Cuttlefish model and how it edges LoRa. He shared the use cases of Cattlefish and who should use it. Rounding up, he gave his advice on how people can get into the machine learning field. He also shared his future research ideas.

On today's episode, we have Daniel Rock, an Assistant Professor of Operations Information and Decisions at the Wharton School of the University of Pennsylvania. Daniel's research focuses on the economics of AI and ML, specifically how digital technologies are changing the economy. Daniel discussed how AI has disrupted the job market in the past years. He also explained that it had created more winners than losers. Daniel spoke about the empirical study he and his coauthors did to quantify the threat LLMs pose to professionals. He shared how they used the O-NET dataset and the BLS occupational employment survey to measure the impact of LLMs on different professions. Using the radiology profession as an example, he listed tasks that LLMs could assume. Daniel broadly highlighted professions that are most and least exposed to LLMs proliferation. He also spoke about the risks of LLMs and his thoughts on implementing policies for regulating LLMs.

We are excited to be joined by J.D. Zamfirescu-Pereira, a Ph.D. student at UC Berkeley. He focuses on the intersection of human-computer interaction (HCI) and artificial intelligence (AI). He joins us to share his work in his paper, Why Johnny can't prompt: how non-AI experts try (and fail) to design LLM prompts. The discussion also explores lessons learned and achievements related to BotDesigner, a tool for creating chat bots.

In this episode, we are joined by Ryan Liu, a Computer Science graduate of Carnegie Mellon University. Ryan will begin his Ph.D. program at Princeton University this fall. His Ph.D. will focus on the intersection of large language models and how humans think. Ryan joins us to discuss his research titled "ReviewerGPT? An Exploratory Study on Using Large Language Models for Paper Reviewing"

The creators of large language models impose restrictions on some of the types of requests one might make of them. LLMs commonly refuse to give advice on committing crimes, producting adult content, or respond with any details about a variety of sensitive subjects. As with any content filtering system, you have false positives and false negatives. Today's interview with Max Reuter and William Schulze discusses their paper "I'm Afraid I Can't Do That: Predicting Prompt Refusal in Black-Box Generative Language Models". In this work, they explore what types of prompts get refused and build a machine learning classifier adept at predicting if a particular prompt will be refused or not.

Our guest today is Maciej Świechowski. Maciej is affiliated with QED Software and QED Games. He has a Ph.D. in Systems Research from the Polish Academy of Sciences. Maciej joins us to discuss findings from his study, Deep Learning and Artificial General Intelligence: Still a Long Way to Go.

Today on the show, we are joined by Lin Zhao and Lu Zhang. Lin is a Senior Research Scientist at United Imaging Intelligence, while Lu is a Ph.D. candidate at the Department of Computer Science and Engineering at the University of Texas. They both shared findings from their work When Brain-inspired AI Meets AGI. Lin and Lu began by discussing the connections between the brain and neural networks. They mentioned the similarities as well as the differences. They also shared whether there is a possibility for solid advancements in neural networks to the point of AGI. They shared how understanding the brain more can help drive robust artificial intelligence systems. Lin and Lu shared how the brain inspired popular machine learning algorithms like transformers. They also shared how AI models can learn alignment from the human brain. They juxtaposed the low energy usage of the brain compared to high-end computers and whether computers can become more energy efficient.

On today's show, we are joined by Michael Timothy Bennett, a Ph.D. student at the Australian National University. Michael's research is centered around Artificial General Intelligence (AGI), specifically the mathematical formalism of AGIs. He joins us to discuss findings from his study, Computable Artificial General Intelligence.

We are joined by Koen Holtman, an independent AI researcher focusing on AI safety. Koen is the Founder of Holtman Systems Research, a research company based in the Netherlands. Koen started the conversation with his take on an AI apocalypse in the coming years. He discussed the obedience problem with AI models and the safe form of obedience. Koen explained the concept of Markov Decision Process (MDP) and how it is used to build machine learning models. Koen spoke about the problem of AGIs not being able to allow changing their utility function after the model is deployed. He shared another alternative approach to solving the problem. He shared how to engineer AGI systems now and in the future safely. He also spoke about how to implement safety layers on AI models. Koen discussed the ultimate goal of a safe AI system and how to check that an AI system is indeed safe. He discussed the intersection between large language Models (LLMs) and MDPs. He shared the key ingredients to scale the current AI implementations.

An assistant professor of Psychology at Harvard University, Tomer Ullman, joins us. Tomer discussed the theory of mind and whether machines can indeed pass it. Using variations of the Sally-Anne test and the Smarties tube test, he explained how LLMs could fail the theory of mind test.

The application of LLMs cuts across various industries. Today, we are joined by Steven Van Vaerenbergh, who discussed the application of AI in mathematics education. He discussed how AI tools have changed the landscape of solving mathematical problems. He also shared LLMs' current strengths and weaknesses in solving math problems.

Fabricio Goes, a Lecturer in Creative Computing at the University of Leicester, joins us today. Fabricio discussed what creativity entails and how to evaluate jokes with LLMs. He specifically shared the process of evaluating jokes with GPT-3 and GPT-4. He concluded with his thoughts on the future of LLMs for creative tasks.

Barry Smith and Jobst Landgrebe, authors of the book "Why Machines will never Rule the World," join us today. They discussed the limitations of AI systems in today's world. They also shared elaborate reasons AI will struggle to attain the level of human intelligence.

While the possibilities with AGI emergence seem great, it also calls for safety concerns. On the show, Vahid Behzadan, an Assistant Professor of Computer Science and Data Science, joins us to discuss the complexities of modeling AGIs to accurately achieve objective functions. He touched on tangent issues such as abstractions during training, the problem of unpredictability, communications among agents, and so on.

Julian Michael, a postdoc at the Center for Data Science, New York University, joins us today. Julian's conversation with Kyle was centered on the NLP community metasurvey: a survey aimed at understanding expert opinions on controversial NLP issues. He shared the process of preparing the survey as well as some shocking results.

Kyle shares his own perspectives on challenges getting insight from surveys. The discussion ranges from commentary on the market research industry to specific advice for detecting disingenuous or fraudulent responses and filtering them from your analysis. Finally, he shares some quick thoughts on the usage of the Chi-Square test for interpreting cross tab results in survey analysis.

Jeff Jones, a Senior Editor at Gallup, joins us today. His conversation with Kyle spanned a range of topics on Gallup's poll creation process. He discussed how Gallup generates unbiased questionnaires, gets respondents, analyzes results, and everything in between.

Gireeja Ranade, a University of California at Berkeley professor, speaks with us today. She presented her study on implementing inclusive study groups at scale and shared the observed student performance improvements after the intervention.

Today, we are joined by David Bourget. David is an Associate Professor in Philosophy at Western University in London, Ontario. David is also the co-director of the PhilPapers Foundation and Director of the Center for Digital Philosophy. He joins us to discuss the PhilPapers Survey project. The PhilPapers survey was initially taken in 2009, but there was a follow-up survey in 2020. David discussed the need for the subsequent survey and what changed. He mentioned the metric for measuring the opinion changes between the 2009 and 2020 surveys. He also shared future plans for the PhilPapers surveys.

Today's show focused on an essential part of surveys — missing values. This is typically caused by a low response rate or non-response from respondents. Yajuan Si is a Research Associate Professor at the Survey Research Center at the University of Michigan. She joins us to discuss dealing with bias from low survey response rates.

We are joined by two guests today, Mariah, a Ph.D. student in the CORE Robotics Lab at Georgia Tech, and Matthew Gombolay, the Director of the CORE Robotics Lab. They both discuss practices for measuring a respondent's perception in a survey.

Ever wondered what your next career would be? Today, Keyon Vafa, a computer science Ph.D. student at Columbia University, joins us to discuss his latest research on developing a machine-learning model for career prediction. Keyon extensively spoke about how the model was developed and the possibilities it brings.

Noura Insolera, a Research Investigator with the Panel Study of Income Dynamics (PSID), joins us to share how PSID conducts longitudinal household surveys. She also shared some interesting findings from their data exploration, particularly on the observation and trends in food insecurity.

Susan Gerbic joins Kyle to review some of the surveys Data Skeptic has launch, draft a new survey about podcast listening habits, and then review the results of that survey. You can see those results at the link below. https://survey.dataskeptic.com/survey/result/1675102237053 Watch the videos Susan mentioned on her Youtube page at the link below. https://www.youtube.com/playlist?list=PL7VAuaQDhPTVaLeI1IcpYph5lH19xA1u4

The use of social bots to fill out online surveys is becoming prevalent. Today, we speak with Sara Bybee, a postdoctoral research scholar at the University of Utah. Sara shares from her research, how she detected social bots, the strategies to curb them, and how underrepresented groups can be more represented in surveys.

Our guest today is Zoltán Kekecs, a Ph.D. holder in Behavioural Science. Zoltán highlights the problem of low replicability in journal papers and illustrates how researchers can better ensure complete replication of their research and findings. He used Bem's experiment as an example, extensively talking about his methodology and results.

On the show, Iñigo Martinez, a Ph.D. student at the University of Navarra shares his survey results which investigated how data practitioners perform data science projects. He revealed the methodologies typically used by data practitioners and the success factors in data science projects.

On the show today, Dino Carpentras, a post-doctoral researcher at the Computational Social Science group at ETH Zürich joins us to discuss how opinion dynamics models are built and validated. He explained how quantifying opinions is complex, and strategies to develop robust models for measuring and predicting public opinions.

Crafting survey questions is one thing but getting your audience to fill it is yet another. On the show today, we speak with Alexander Nolte, an Associate Professor at the University of Tartu. Alexander discussed the use of Casual Affective Triggers (CAT) to incentivize people to accept survey invitations and improve the completion rate. He revealed the impact of CATs on survey response rates from a study he conducted.

Traditional surveys have straight-jacket questions to be answered, thus restricting the information that can be gotten. Today, Ziang Xiao, a Postdoc Researcher in the FATE group at Microsoft Research Montréal, talks about conversational surveys, a type of survey that asks questions based on preceding answers. He discussed the benefits of conversational surveys and some of the challenges it poses.

Today, Jenny Tang, a Ph.D. student of societal computing at Carnegie Mellon University discusses her work on the generalization of privacy and security surveys on platforms such as Amazon MTurk and Prolific. Jenny shared the drawbacks of using such online platforms, the discrepancies observed about the samples drawn, and key insights from her results.

This episode kicks off the new season of the show, Data Skeptic: Surveys. Linhda rejoins the show for a conversation with Kyle about her experience taking surveys and what questions she has for the season. Lastly, Kyle announces the launch of survey.dataskeptic.com, a new site we're launching to gather your opinions. Please take a moment and share your thoughts!

It may be intuitive to think crowdfunding a project drives its innovation and novelty, but there are no empirical studies that prove this. On the show, Johannes Wachs shares his research that sought to determine whether crowdfunding truly drives innovation. He used board games as a case study and shared the results he found.

There were reports of Russia's interference in the 2016 US elections. In today's episode, Koustuv Saha, a researcher at Microsoft Research walks us through the effect of targeted ads for political campaigns. Using practical examples, he discusses how targeted ads can propagate fake news, its ripple effects on electioneering, and how to find a sweet spot with targeted ads.

There is an unsung kind of ad fraud brewing in the ad tech space — placement laundering fraud. On the show, Jeff Kline discusses what placement laundering fraud is, how it can be identified, and possible solutions to it. Listen to learn more.

Bosko Milekic, the Co-founder of Optable, a data collaboration platform for the media and advertising industry, joins us today. Bosko talked about the clean rooms, the technology driving data privacy during collaboration. He discussed why clean rooms are gaining widespread adoption, and how users can exploit Optable's clean room platform for a secured data-sharing experience.

Kerstin Bongard-Blanchy is a Research Associate at the University of Luxembourg. She joins us to discuss her study that investigated dark patterns in web designs. She discussed the results, the effect of dark patterns effect on users, whether an average user can detect them, and the way forward to a more ethical web space.

We are joined by Anthony Katsur, the CEO of IAB Tech Lab. Anthony discusses standards within the ad tech industry. He explained how IAB Tech Lab set and propagates global standards, actions to ensure compliance from advertisers, and industry trends for a more privacy-centric ad tech space.

When we navigate a webpage, it is fairly easy for our mouse movement to be tracked and collected. Today, Luis Leiva, a Professor of Computer Science discusses how these mouse tracking data can be used to predict age, gender and user attention. He also discusses the privacy concerns with mouse tracking data and possible ways it can be curtailed.

On the show, Aleksandra Urman and Mykola Makhortykh join us to discuss their work on the comparative analysis of web search behavior using web tracking data. They shared interesting results from their analysis, bordering around the user preferences for search engines, demographic patterns, and differences between how men and women surf the net.

Did Aristotle Use a Laptop? That's a question from the StrategyQA benchmark which highlights the stretch goals for current artificial intelligence systems. Answering a question like that requires several cognitive steps and reasoning. Constructing a dataset of similarly challenging questions is a major undertaking. On today's episode, Mor Geva returns to share details about the creation of StrategyQA and the larger Big Bench dataset it has been included in.

While at first glance, the use of ad blockers drops the revenue of news publishers, this may not be completely true. On the show today, Shunyao Yan, an Assistant Professor in Marketing at Leavey School of Business, Santa Clara University, discussed the effect of ad blockers on news consumption and how ad blockers can potentially be helpful for news publishers.

People who do not want their data tracked and shared online can pay a token for a cookie paywall. But are the websites keeping to their side of the bargain? Victor Morel, a Postdoc candidate at the Chalmers University of Technology joins us to discuss his work around auditing the activities of cookie paywalls. He discussed the findings from his analysis and proffers some solutions to making cookie paywalls more transparent.

The advancement of generative language models has been a force for good, but also for evil. On the show, Avisha Das, a post-doctoral scholar at the University of Texas Health Center, joins us to discuss how attackers use machine learning to create unsuspecting phishing emails. She also discussed how she used RNN for automated email generation, with the goal of defeating statistical detectors.

Peter Gloor, a Research Scientist at the MIT Center for Collective Intelligence, takes us on a new world of tribe classification. He extensively discussed the need for such classification on the internet and how he built a machine learning model that does it. Listen to find out more!

We hear about the impeccable achievements of GPT-3 models, but such large generative models come with their bias. On the show today, Conrad Borchers, a Ph.D. student in Human-Computer Interaction, joins us to discuss the bias in GPT-3 for job ads and how such large models can be de-biased. Listen to learn more!