Have AI initiatives added measurable value to your D&A function? How do you demonstrate the value and impact of AI initiatives on the overall data analytics process within your organization?
Sort By:
Oldest
Chief Data Officer in Government5 months ago
We are still investigating this topic as we work with analytical models and etc on a regular basis and where is the delineation of analytics vs AI. Many of the "AI" initiatives I have reviewed are just 'regular analytics' we would be doing every day.Sr. Director, GenAI Program Management in Healthcare and Biotech5 months ago
Productivity metrics are fairly straight forward to measure and have had a positive influence within many D&A functions. For instance, Github Copilot or ChatGPT is very helpful with resolving coding gaps - accelerating model development takt time. Through A/B like hackathons amongst peer developer communities, or simple surveys you can measure benefits and solve business problems getting two birds with one stone.Chief Data Officer in Software5 months ago
Value measurements through AI are the same as any D&A initiative. They either: promote improved decision making; increase business and process efficiencies (and productivity); or the help to manage or mitigate risk. To better understand how to to measure the value of these things, I would recommend the book 'Infonomics' by Doug Laney. VP of Data and Analytics in Education5 months ago
It may depend on how you define AI. Are you specifically referring to GenAI? If so, then we aren't all the way to completion and measurement yet. If AI is inclusive of ML, then yes we have measurable value. We've taken an approach around prioritization that usually leads to value measures after completion being straightforward. For example, in prioritizing an AI effort, we use the following inputs:1-yr cost of delay
Non-monetary business impact
Calendar time-to-implementation (TTI)
"Billable" labor hours required
Probability of a successful production implementation
Indicator for whether the project supports an enterprise-wide initiative
We weigh the cost of delay (lost revenue, missed savings, etc) more heavily than the non-monetary impact, but both are important. We've had multiple successful deliveries ranging from generating $10M+ in additional ongoing annual revenue to reducing cost to produce a product/service to increases in lives saved. Once a product or service is in production, we then measure whether the value estimates (both monetary and non-monetary) were accurate and use that to not only learn but also show value to the business.
Director of Data5 months ago
We leverage the CRISP-DM methodology which defines the business problem statement, business success criteria, business success metrics. In the modeling and evaluation phases of CRISP-DM the proposed solution is tested to see if it meets the objectives. Not only satisfying the business problem, but also the success criteria, and success metrics. In this way it's clear if the effort will provide the desired results or not.