Alternative Data Weekly #153
Alternative Data Weekly #153
September 29, 2023
Thanks for being here!
Alternative Data Weekly is Powered by Vertical Knowledge
Announcements:
Vertical Knowledge has a team in London for the Eagle Alpha Conference October 12. Let me know if you will be in London for this event.
Theme that emerged in this week’s email is … data is core to any AI model. Applying this tech to solving real world problems is the key.
QUOTES
“Data-driven innovation offers enormous potential in many sectors, including agriculture, education, energy, health care, public safety, transportation, and so much more.” – Center for Data Innovation
“Data cannot be an afterthought in generative AI. Rather, it is the core fuel that powers the ability of a business to capture value from generative AI.” - McKinsey
News Articles
Podcasts
Cool Charts
Final Thoughts (Networking)
#1 – Gillian Diebold of the Center of Data Innovation published Overcoming Barriers to Data Sharing. September 2023.
My Take: The world has changed in the past 10-15 years. The “datafication” of our society is happening and happening quickly. We are at the very early stages of putting this data to use (#dataforgood). There are of course technical challenges, but also legal issues, standards that need to be created, plus the social impact of simply more information being in the public sphere.
Organizations like the Center for Data Innovation are leading the way when it come to ensuring data is collected, processed, and shared in a way that creates the most benefit individuals, organizations, and society at large. This is important work, because the change is happening…that won’t be stopped.
#2 – A team of authors from McKinsey published The data dividend: Fueling generative AI. September 2023.
My Take: There is a lot of noise in the data world. AI & LLMs have everyone excited, and rightfully so…if you’ve used the early tools, they are cool. But there has yet to a be that killer data app or data’s “iPhone moment” where it becomes obvious how value will be created. This article shares a great framework from which a CDO can begin to organize a data/AI strategy.
See below for some cool charts from this same article.
#3 – Gradient Flow published 7 Must-Have Features for Crafting Custom LLMs. September 2023.
My Take: Taking this amazing Generative AI / LLM technology and applying it to the real world. That is the goal. This article reminds us that this isn’t about technical finesse, but rather aligning the tech with real-world applications.
Here are the 7 features:
Versatile and Adaptive Tuning Toolkit
Human-Integrated Customization
Data Augmentation and Synthesis
Facilitation of Experimentation
Distributed Computing Accelerator
Unified Lineage and Collaboration Suite
Excellence in Documentation and Testing
BONUS 1: Abraham Thomas authored The Worst Outcome is a Mediocre Success. September 2023. “make sure that there’s a well-defined distinction between success and failure. Don't fall in the messy middle.”
What else I am reading:
The FTC published Remarks of BCP Director Samuel Levine at 2023 Consumer Data Industry Association Law & Industry Conference. September 2023.
Benn Stancil’s So you want to sell to the enterprise?. September 2023.
Ethan Mollick’s Everyone is above average. September 2023.
Random Walk’s Stretching SAAS sales cycles. September 2023.
Steven Jacobs from Bobsled published Why LinkUp is Doubling Down on Data Sharing. September 2023.
Quinn Emmanuel published The Legal Landscape of Web Scraping. April 2023.
Source: The Joe Reis Show interviewed Juan Sequeda - The Power of Knowledge Graphs and LLMs on Structured Data in the Enterprise. September 2023.
My Take: I am not sure how this is the first Joe Reis podcast I’ve highlighted. This is a great library of interviews. Honest. No BS. Not sale-sy.
What are the tasks you have and how can generative AI help make you more productive. Everyone at Juan’s company uses generate AI in their jobs.
To do anything with AI, you’ve got to have the data. There is a ton of noise … how do we separate hype from the pragmatic stuff that is actually working.
We are going from “data first world” to “knowledge first world”.
“AI is not going to take your job away … another person using AI to do that job better will take your job away.” (minute 42:00)
Highlights (51-minute run time):
Minute 00:30 – intro
Minute 06:00 – is there anything to talk about besides AI? Structured / unstructured data
Minute 07:45 – discussion of Juan’s current work; combining LLMs with structured data
Minute 11:00 – how much does a knowledge graph improve the accuracy of LLM systems translating to SQL
Minute 16:00 – complexity of questions & complexity of schemas
Minute 21:00 – chat interfaces (language to code) will be a feature
Minute 22:00 – back to the beginning of the web; LLMs is a web moment (web vs internet)
Minute 29:00 – google search is all done through knowledge graphs (“things not strings”)
Minute 31:00 – thoughts on ChatGPT enterprise (“if you are not using generative AI in your organization, you are an idiot”)
Minute 34:30 – get legal & compliance teams on the data journey from the beginning
Minute 39:30 – we are still very early on the data / LLM / generative AI world; predictions
Minute 42:00 – AI is not going to take your job away … another person using AI to do that job better will take your job away
Minute 46:00 – the power of knowledge graphs; unifying LLMs & knowledge graphs
Bonus: I enjoyed this Lex Fridman conversation with Walter Isaacson.
Source: McKinsey’s The data dividend: Fueling generative AI. September 2023.
My take: The CDO should also prioritize initiatives that can provide the broadest benefits to the business, rather than simply support individual use cases. Get some early wins too!
One more …. Ensuring data quality in web scraping projects. September 2023.
Source: Networking in Person
When this email is published (Friday 7am ET … Sept 29th), I will have just spent the prior day at the Sept 28th Neudata conference in SF.
Interestingly, this same San Francisco conference February 27, 2020 was the final trip I took prior to the covid lockdowns. I can still remember being at a crowded Ash Wednesday (Feb 26th) mass and first noticing people were tentative about being in such a large crowd. A couple weeks later we were all locked down.
I am happy to be back at it. These types of events are important for networking, brainstorming, meeting prospective clients … & listening to them talk about their businesses.
Virtual meetings are great and save a ton of time & money, but there is benefit to being in person, seeing body language, and (most importantly) getting those impromptu serendipitous meetings that just might make someone’s day.
Scott Galloway is multiples more eloquent & opinionated than me, but I agree with his sentiment about putting yourself out there. It gets easier the more you try it.