Halloween Data Spooktacular: What Keeps Data Leaders Up at Night?
It is the season of spooks and specters. At Corinium, our thoughts turn to the things that haunt us... in the world of data and analytics, of course
In the spirit of the season (get used to the puns, it’s a special occasion) we couldn't resist the temptation to dig into the darker side of data and analytics.
We've conjured up insights from some of the industry’s leading minds to find out just what sends chills down their spines. Grab your favorite candy and a cup of hot cocoa and let’s unwrap these terrifying tales together!
Voices Beyond the Grave: The LLM Output Omen
Louis DiModugno, the Bowtie Data Guy and Managing Partner at Data Curiosity quivers at the thought of our growing reliance on large language models (LLMs) like ChatGPT.
“The ease of submitting a question to ChatGPT and the authoritative response that comes back is truly as frightening as 'a voice beyond the grave’,” he warns.
Just as ghosts might deceive the living with their whispers, relying too heavily on LLMs without proper scrutiny could lead businesses astray. As enchanting as these automated answers might be, DiModugno urges us to tread carefully, lest we fall under their spell without questioning the sources behind the sorcery.
The Sorcerer’s Dilemma: Demystifying Data and Technologies
Cecilia Dones, Founder and Principal at data and analytics consultancy 3 Standard Deviations, worries that individuals wielding the almost incomprehensible power of data and analytics might be mistaken for witches or sorcerers (instead of the superheroes they so clearly are...)
“Data is so complex and mystical in some ways that stakeholders might view me as a magician or sorcerer, like Dr. Strange from the Marvel movies.”
She wants to clarify: “I am not as good as an actor as Benedict Cumberbatch.”
On a more serious note, she also highlights the danger of rushing to capitalize on data and technologies without fully understanding them, fearing this could create confusion, complexity, and missed opportunities.
“In the effort to capitalize quickly on the opportunities that data and technologies can bring, as leaders we can inadvertently create more confusion and complexity about the industry and unintentionally miss out on creating value for the consumer and the business.”
A truly terrifying thought for any enterprise!
The Beast has Two AIs: The Importance of Data Readiness
Asha Saxena, Founder and CEO of Women Leaders in Data and Author of The AI Factor, directs her baleful eye at our lopsided love affair with AI.
“I'm genuinely concerned about the prevailing fascination with AI, yet there is an insufficient allocation of resources towards data management and integrity,” she says. “AI has become the latest shiny object in the tech world, and people tend to forget a fundamental truth: if we don't meticulously prepare our data, it significantly impacts the quality of the AI model's output and results.”
In her book, ‘The AI Factor,’ she introduces a comprehensive data readiness framework to help practitioners ensure their data is up to snuff before letting AI loose. Saxena emphasizes the importance of data quality, governance, preprocessing, and the need for raw data to be cleaned and enriched.
Neglecting these crucial steps is like playing with an Ouija board - you might not like what you summon!
Beware the Data Demons: Prioritizing Quality at the Point of Collection
Onyinye Enyia Daniel, PhD, Vice President of Data and Analytics Strategy at Highmark Health warns of the nightmares conjured by using bad or incomplete upstream data in downstream machine learning models and algorithms.
“Organizations should commit to upstream data due diligence and quality at the point of data collection,” she insists, highlighting the need for robust and fair algorithmic outputs. “Bad or incomplete upstream data is the stuff of nightmares!”
Ignoring data quality upstream is akin to inviting a vampire into your home - once you let it in, it’s only a matter of time till you feel the bite…
It’s ALIVE! Beware Data Management Franken-Structures
Finally, DiModugno is mighty afeared about the ‘Frankenstein’ structures of some data management environments.
He critiques large software companies for their ‘one solution fits all’ mentality, resulting in a mishmash of tools and platforms stitched together like Dr. Frankenstein’s monster.
“A number of large software companies have a ‘one solution for many issues’ approach and try to complete an end-to-end offering,” he says. “But in my opinion, it's the Frankenstein approach. This may give you an optimal solution, but you need to make sure it doesn’t take on a life of its own (It’s ALIVE!).”
As the leaves fall and the nights grow longer, these data leaders remind us to stay vigilant, questioning the ghosts in our machines and ensuring our data practices are more treat than trick.
Happy Halloween, and may your data be ever in your favor!
___
Hear more from our global network of data and analytics innovators by downloading Corinium's Top 100 Global Innovators in Data and Analytics 2023 here
