Organizations deploy artificial intelligence and machine language with trained models that continuously process data to arrive at new results – but what happens if you don’t follow these AI systems to make sure they stay relevant?
In 2021 Spiceworks reported: questionnaire results that revealed: “Nearly a third (31%) of professionals surveyed said their organizations are now using artificial intelligence (AI), and 43% are researching the technology. About 34% reported that their companies had not implemented AI projects.”
This and other studies show that most companies are in the early stages of AI adoption – and most likely have not yet thought about change management for their AI systems and what it takes to get their AI systems up, running and to keep it relevant.
TO SEE: Ethical Policy for Artificial Intelligence (Tech Republic Premium)
How important is an AI maintenance and tuning strategy?
In 2016, Microsoft developed a chatbot called Tay. Tay is designed to learn from human interactions on social media. Shortly after it was deployed, Tay started learning from social media and started spitting it out racist and hateful comments. Tay was one of the first forays into AI learning models and social media, and it ended in disaster.
The Tay incident is not an isolated incident.
AI systems can easily become damaged and lose their effectiveness by processing “toxic” data (such as deepfakes) maliciously injected into the data the AI learns from. AI systems can also process data from a limited number of sources that begin to lose effectiveness over time as newer, more relevant data sources come online.
“Risks after the implementation of AI arise from an incomplete understanding of the behavioral boundaries, untested failure modes, and susceptibility to manipulation by hostile elements in the deployment environment,” said Francois Candelon, Global Director of the BCG Henderson Institute. As data and environments change, AI systems must grow and adapt.
The question is, how many organizations have thought about how they are going to verify, maintain and tune AI systems so that they remain relevant?
Ways to ensure AI systems keep their function
Here are five strategies companies can use to ensure their AI stays relevant and doesn’t lose the value of their AI investment over time:
1. Engage a diverse team of AI reviewers
“To create effective AI+Human systems, leaders need to engage developers, managers, users, consumers, and others to understand the application context of AI,” Candelon said.
Without a diverse set of AI workers, you risk missing out on important perspectives and elements for your AI. This diverse team needs to be kept together as your AI system evolves so that your system can maintain its effectiveness and relevance.
If building a team of AI architects has been a challenge for your organization, the experts at TechRepublic Premium has a rental package that should help. It includes tools to find and hire the right people for the team.
2. Verify AI accuracy with external benchmarks
If AI predictions start to deviate significantly from what the organization has experienced in the past, questions need to be asked:
- Have circumstances really changed?
- Is there a data source or human input that is not included in the AI data store?
- Are the right algorithms and questions formulated?
- If the physical reality of what is actually happening (e.g. critical path equipment on production lines) doesn’t match what the AI predicts will fail, does the AI need to be recalibrated?
3. View and Update Data Sources
Are there new data sources now available that are more comprehensive and accurate that were not available when the AI system was first implemented? If so, it makes sense for IT and other AI contributors to include this data so that the AI works with as complete a set of data as possible.
4. Evaluate whether business use cases have drifted
The business use case your AI was originally built for was focused on customer acquisition, but times (and the business case) have changed and now the focus is on customer retention.
When a business use case for AI changes, the AI must change with it, otherwise the AI will become obsolete and you may lose the benefit of your time, effort and monetary investment.
5. Evaluate for risk
If your AI analyzes human interactions on social media for the purpose of responding to or extracting information, it has a duty to evaluate AI processing for potential risks.
“Deep stakeholder engagement throughout the AI lifecycle will help uncover blind spots and ensure effective monitoring of risk in data, modeling, tradeoffs and concept drive,” said Caldelon.
So do effective governance and change management best practices as AI systems continue to evolve and learn from the data they process.