Alternative Data Weekly #161

November 24, 2023

Thanks for being here!

---

The Alternative Data Weekly is Powered by Vertical Knowledge

---

Announcement(s):

Vertical Knowledge will have a team at Neudata’s December 7^th conference in NY.  Let me know if you plan to be in attendance. 

---

Theme that emerged in this week’s email is … asking better questions is the key to better decisions.

QUOTES

“We think models are beginning an age of “intelligence-as-a-service”, or IQaaS, and the winners will be decided in a race for talent, data, and compute power.” – Coatue’s AI: The Coming Revolution

• News Articles

• Podcasts

• Cool Charts

• Final Thoughts (Types of Podcasts)

  ---

#1 – QuantMinds published Quant in flux – QuantMinds eMagazine 2023.  November 2023.

• My Take: This article might take a few steps to access, but interesting & worth it.  The discussion of new trends in the quant industry, including ML & GenAI, Sentiment analysis, LLMs, Liquidity risk.  The general theme is the challenge of incorporating these new technologies into the day-to-day decision-making process.

  ---

#2 – Jason Derise’s The Data Score published Questions First, Data Second.  November 2023.

• My Take: Ask better questions.  Domain expertise is extremely important.  Once you have the right question, you can then move on to the data.  I love the reference to the Bayesian approach in terms we can understand.

“The most serious mistakes are not being made as a result of wrong answers. The true dangerous thing is asking the wrong question.” - Peter Drucker

#3 – Exabel published a white paper How Discretionary Managers Combine Alternative Datasets to Discover Deeper Insights.  November 2023.

• My Take: Combining datasets is the next step in the evolution of using data to improve decision-making.  This is hard to do.  Tools are improving.  Once more accessible, the combinations become endless and the importance of asking the right questions (“prompt engineering”) & understanding which datasets, when combined, create the most robust corpus of information for your use case, will become extremely valuable.

---

What else I am reading:

Kate Strachnyi’s DATAcated 2024 Trends & Predictions.  November 2023.

Timothy B Lee’s Kyle Vogt exits Cruise, leaving Waymo as the clear robotaxi leader.  November 2023.  (my take: self-driving cards will take forever and then happen all at once ... changing the world for the better).

Ethan Mollick’s Almost an Agent: What GPTs can do.  November 2023.

Pre-order Ethan Mollick’s forthcoming book: Co-Intelligence.

---

Source: Striim’s What’s New in Data Podcast published an interview with John K Thompson, Global Head of Artificial Intelligence (AI) at EY.  Causal Artificial Intelligence, potential AI pitfalls, getting executive buy-in.  November 2023.

• My Take: John is co-author of "Causal Artificial Intelligence: The Next Step in Effective Business AI".  He discusses “understanding why” as the driving force behind the book; he also touches on his writing process.

• Bottom line, generative AI allows you to ask better questions.  This is not just a search engine on steroids.  The real question is how creative can you be (minute 11:45)?

• AI benefits from shared ownership… the data teams need the analytics team … the analytics team need the subject matter experts … the execs need all to work together.

• John discusses common pitfalls including, 1- not getting executive buy-in; 2 - inaccurate or out-of-date content; and, 3- for causal AI in particular, the software is catching up (not yet easy to use).

---

Highlights (32-minute run time):

• Minute 00:30 – interview starts; John’s background; discussion of his book

• Minute 04:00 – the story behind writing the book & the writing process

• Minute 07:30 – three pillars of AI

• Minute 09:30 – how does generative AI get the context?

• Minute 10:30 – examples of John’s interactions with GPT (ask better questions)

• Minute 12:30 – preventing the incorrect outcomes (RAG – retrieval augment generation)

• Minute 14:30 – how RAG works

• Minute 16:30 – approach to building our production AI (John is an implementer)

• Minute 18:10 – pitfalls and how to avoid

• Minute 21:00 – how to test model fit and QA-ing output

• Minute 24:00 – COVID impact on models

• Minute 26:00 – who should own AI from an implementation perspective?

• Minute 28:00 – how should enterprise think about proprietary data use in AI models

• Minute 30:30 – importance of bringing it back to metrics that business-users can understand

---

BONUS: Mapscaping podcast interviewed Auren Hoffman Entity Resolution with Placekey (40 minutes).

---

BONUS 2: Thanksgiving tradition is to listen to this one: Arlo Guthrie’s Alice’s Restaurant.

---

Source: Coatue’s The AI Revolution.  November 2023.

Too many good charts in this one … please check out the full 115-page deck.

---

---

---

---

---

Types of podcast …

1-     ENTERTAINING

• Bill Simmons’ BS Podcast

• Spittin’ Chicklets

• Most things by Evergreen Podcasts

---

2-     LEARNING

• News (WSJ, NYT, Economist)

• Free Press

• Prof G

• Founders

• Acquired

• Depends on interviewee:

  • Tim Ferriss

  • Lex Fridman

  • Up2

  • Master’s in Business

  • Joe Rogan

  • Lenny’s Podcast

    ---

3-     REALLY LEARNING (sitting down, taking notes, pausing occasionally)

• The Riff (new one … check it out)

• World of DAAS

• Hedgineer Podcast

• Long list of data-related podcasts that are consistently named in the ADW

  • I See Data People (Omri & Evan)

  • Alternative Data Podcast (Mark Fleming-Williams)

  • The Data Cloud (Snowflake)

  • Data Radicals (Alation)

  • The Data Couch (Louise de Leyritz)

  • The Data Engineering Podcast (Tobias Macey)

  • The Data Exchange (Ben Lorica)

  • Data Framed Podcast (Datacamp)

  • What’s New in Data (Striim)

  • The Data Chief (Cindi Howson)

  • The Analytics Engineering Podcast (dbt)

  • Oxycast (Oxylabs)

---

Name a couple I am missing!