Conference season is in full swing. Industry events and user conferences are bringing together practitioners and thought leaders to discuss the trends shaping tomorrow’s markets, and no other topic is more prevalent than the potential game-changing impact of AI (artificial intelligence).
It’s impossible to escape the buzz. AI’s influence is undeniable. But to what end?
I’ve seen three reactions at these events:
- Give Me More: Early adopters who love seeing the latest trends
- Tell Me How To Impact Business: Rationalists who want to cut through the hype
- Stop Talking About AI As The Answer To Everything: Pessimists are skeptical
I am a pragmatist, and most executives evaluating AI are as well. They want to know how AI fits into their business and what is needed to realize value. AI is not new, but recent advancements now make it viable, and this will produce new, yet-to-be-envisioned models. As Alvin Toffler once said, “Technology feeds on itself. Technology makes more technology possible.”
AI IS DEPENDENT ON DATA
In my experience, it all starts with data. AI is entirely dependent on big data, and data is required to train LLM models that fuel AI generative models. Without lots of data—the correct data—AI falls flat.
There is a term for this in computer science: garbage in, garbage out (GIGO), which is the concept that flawed, biased, or poor-quality (“garbage”) information produces a result of similar (“garbage”) quality.
Clean data helps models learn effectively and allows them to understand patterns and relationships. Good data quality is the extent to which any dataset is accurate, complete, valid, consistent, and timely within an organization. Data integrity is data’s reliability, consistency, and accuracy over time. Poor data quality means AI models create inaccurate predictions or advice. This can lead to faulty decisions or biases that could have negative consequences.
Consider simple use cases in healthcare or finance.
- Healthcare: Faulty patient data surrounding comorbidities could lead to harmful health guidance. AI does not have context for these extenuating circumstances.
- Finance: Poor data quality can result in inadequate financial advice, creating trust issues and financial liability from a fiduciary.
Poor data quality and integrity compounded with data silos, lack of integration, and a skills gap make the problem more profound. According to an IBM study, the financial impact of poor data quality on the U.S. economy is estimated to be $3.1 trillion annually, and only 53% of companies surveyed can leverage big data for a competitive advantage. AI will highlight any decencies in the overall data model and schema. AI won’t function as promised if the data isn’t accurate.
DATA IS THE PROBLEM AND SOLUTION FOR AI
The proper use of AI depends on data. There are steps to take to get your data house in order and make the most of AI advancements.
Begin with an audit of all your company’s data assets. Which ones? Well, all those that will influence AI, such as customer, operational, financial, and sales and marketing data. Consistency, completeness, and sources are evaluated for each data type. The goal is to analyze the root causes of data problems and initiate quality and integrity improvements before jumping into AI modeling. Focus on identifying and eliminating the root causes to improve how new data is created. This makes the mitigation efforts much more palatable to correct without a massive clean-up effort that may take months to complete.
This time well spent will provide a solid foundation on which AI efforts can shine, and you can begin testing with simple AI use cases.
CLEAN DATA LEADS TO AI IMPACT
The complexities of AI are vast and comprise the multidisciplinary fields of computer science, mathematics, cognitive science, neuroscience, and engineering. It involves various approaches, including machine learning, deep learning, natural language processing (NLP), and computing power. As Demis Hassabis, cofounder of DeepMind, remarks, AI is “the science of making machines smart.” Clean data ensures AI has the right inputs to create those intelligent machines.
AI platforms such as ChatGPT and CoPilot will undoubtedly help information workers find efficiencies in routine workflows and redundant tasks. Still, for AI to transform an industry, it will require harnessing computing power—the efficiency and speed with which the “machine” can perform complex tasks and handle large amounts of data.
THE FUTURE IS BRIGHT IF THE DATA IS RIGHT
Imagine the future of AI with clean, pristine data where multiple LLMs can be stitched together. Overnight, companies will move to fully autonomous applications, enhancing humans’ capabilities by performing intelligent tasks and solving complex problems. It’s exciting to think of what’s possible. Transforming starts with data quality and integrity. The hard truth today is data quality and integrity are the laggards.