This quarter I’m taking Professor Candogan’s Frontiers in AI for Operational Decision-Making (BUSN 40915). We swept through the fundamentals of AI for the first two weeks. My opinion may be biased because I am already preloaded with all the AI terminologies, but the course infused extreme clarity into my understanding. The following are key slides that I found informative/inspiring. Courtesy to Professor Candogan — I don’t own copyrights! All mistakes are on my own. ...

Academic industry is a beautiful ecosystem that has contributed a lot to human. It’s more easy to criticize something that is supposed to be perfect and pure. This is an article based on paper The junkification of research by Martina Linnenluecke and Carl Rhodes: The 5 stages of the ‘enshittification’ of academic publishing Martina Linnenluecke, Carl Rhodes | The Conversation | January 5, 2026 The social media platforms, e-commerce sites and search engines they were using had noticeably deteriorated in quality. Many had begun to prioritise content from advertisers and other third parties. Profit became the main goal… Enshittification isn’t just confined to the online world. In fact, it’s now visible in academic publishing and occurs in five stages. The same forces that hollow out digital platforms are shaping how a lot of research is produced, reviewed and published. ...

If I need to get something serious done with the help of AI, I would first open an (auxiliary) AI window to let the AI generate my prompts. One good method is to use the Context-Role-Input-Specification-Process-Evaluation (CRISPE) framework. For example You are an experienced [ROLE/DOMAIN EXPERT]. I will provide a high-level or ambiguous description of a task or problem. Your job is to design a high-quality prompt that would enable an AI to effectively accomplish the task or solve the problem. The resulting prompt should be concise, precise, and action-oriented. You may structure it using the CRISPE framework (Context, Role, Instructions, Scope, Parameters, Examples) or any relevant subset of it, if helpful. Before presenting the final prompt, ask me exactly three high-impact clarification questions that would materially improve the quality and specificity of the prompt. Finally, you will output prompt in a copytable codebox. Works every time. ...

Notes from behavior seminar at Booth. Mental Models and Financial Forecasts Bastianello, Décaire and Guenzel (2025) | SSRN https://dx.doi.org/10.2139/ssrn.5004839 We uncover the mental models financial professionals use to explain their quantitative forecasts, and show how they shape beliefs and return predictability. Using the near-universe of 2.1 million equity analyst reports, we collect the valuation methods analysts adopt to compute their price targets, together with their reasoning, measured as attention to topics, and their associated valuation channels, time horizons, and sentiments. To validate the reliability of our output, we introduce a multi-step LLM prompting strategy and new diagnostic tools. Consistent with a model of top-down and bottom-up attention, we then uncover three sets of facts. First, analysts’ mental models are sparse and rigid, and the choice of attention allocation and valuation methods are jointly determined by both analyst- and firm-characteristics. Second, analysts’ reasoning translates into their quantitative forecasts. Both attention and valuation methods contribute to differences in valuations over time and across analysts, but variation in attention plays a bigger role. Third, we study the extent to which different topics contribute to over and underreaction to information, and show how biases in analysts’ reasoning are reflected in asset prices. Analysts underreact to macroeconomic topics, and overreact to firm-related topics, and this contributes to return predictability. ...

Happy Sunday! Unlike sophisticated, beautifully depressed Brahms or Schubert impromptus, here’s a bit of a jolly music: Whoever came up with this project deserves the highest honor of classical manager. You will find the best resolution available on Medici tv. The hero was Christopher Nupen: Program Note from Medici TV: On August 30, 1969, five young musicians—all well on their way to establishing themselves as renowned international artists—came together to play Schubert’s “Trout” Quintet in London’s new Queen Elizabeth Hall. Their names: Daniel Barenboim, Itzhak Perlman, Pinchas Zukerman, Jacqueline du Pré, and Zubin Mehta. ...

Fun fact: Franz Schubert’s “D” number refers to the Deutsch catalogue, a chronological listing of all his works compiled by musicologist Otto Erich Deutsch (1883–1967). Mainly because Because Schubert lived a short life and wrote many works that were not published immediately, Opus numbers are insufficient/misleading for chronological ordering. Let’s exercise a level of aesthetic pickiness that earlier listeners would have considered an unimaginable luxury—pick your favorite: Ah, Zimmerman got a good photographer for this album cover. ...

Hmmm. So we can trust nothing from the internet now? Debunking the AI food delivery hoax that fooled Reddit Casey Newton Jan 5, 2026 | Platformer | Link A “whistleblower” tried to corroborate his viral post with AI-generated evidence. This is how I caught him. PLUS: Grok’s image-generation crisis, and the rapture over Claude Opus 4.5 A journalist contacted the Reddit whistleblower and tries to get an interview and more information. The whistleblower throws the whistleblower a “fake” ArXiv paper and used a Uber-Eat ID card generated by AI. ...

Brahms wrote it when he was 20. The melodies are warm and expressive (Brahms…). But the music is long and structurally loose. In his fifties (1889) Brahms revised it. The later version has leaner piano, the motives are more rigorous and the overall architecture of the piece is clearer — just a lot more better in all perspectives. The 2nd version is the one often performed today. The piece is not technically demanding but actually (and well, naturally for Brahms) takes a lot of maturity to play well. You need control, patience, emotional restraint to present the textures, tempos and development in the music. ...

It was out a while ago. But reading it is still, well, astounding: Nature’s 10 A fired public-health official, a mosquito breeder and a baby with a smile seen around the world. These are just a few of the remarkable people chosen for Nature’s 10. The Nature’s 10 list recognizes important scientific trends and discoveries over the course of a year and tells the stories of the people involved. It is compiled by Nature’s editors to highlight some of the most influential research and important developments shaping our world. This year’s selection acknowledges scientists peering into the farthest reaches of the Universe and the deepest depths of the ocean, as well as a civil servant who stood up for evidence-based public-health policy amid a vast upheaval in the United States. ...

Chinese are always serious and obsessed about food. On the 26th birthday of Alibaba (2025 Sept 10), AMap released a new Local Discovery Ranking (扫街榜) — for restaurant and general local service recommendations. Unlike traditional online rating platforms rely mainly on user ratings, AMap uses behavioral data including navigation usage, in-store visits, repeat visits, and intentional destination trips to make the rankings more authentic, trustworthy and immune to score manipulation. AMap’s CEO Ning Guo (郭宁) pledge that the rankings would “never be commercialized”: ...