At At The Advancing in Women In Energy’s (AWE) recent Lunch & Learn, our General Manager Liana Jo Ault led a timely discussion on “AI in the Power Industry – The Good, The Bad, The Ugly.” Her talk reminded us that AI is a technology with both extraordinary potential and serious risks if misapplied.
How AI is Being Adopted in the Power Industry Today
While utilities are sometimes seen as slower adopters of new technologies, AI has already found its way into the sector. Current applications include customer service assistants, robotic process automation, spam filters, cybersecurity tools, and—most notably—early use in digital twin modeling for forecasting and planning.
The Good
As Liana put it:
“I think there’s a lot of things that are really good about AI.”
AI can automate mundane tasks, accelerate problem-solving, and provide critical decision support in complex scenarios. Within the power industry, some of the most promising uses include:
- Frequency & Phase Management – Streaming data from PMUs and grid assets can detect and correct shifts caused by outages or demand spikes, enabling automated fast-response reserves.
- Energy Market Forecasting – AI models can combine historical market data, weather forecasts, and asset performance to improve demand prediction and support flexible bidding strategies - paving the way for micro- and macro-level time-of-day markets.
- Operational Safety & Security – Live-stream video and AI pattern recognition can help spot work hazards, while networked alarms can correlate cyber and physical events to strengthen resilience.
- Power Reliability & Maintenance – AI enables operators to view assets holistically - people, communications, and infrastructure. Scenario-based planning can coordinate outages, upgrades, and workforce deployment for optimal reliability.
And of course, as Liana highlighted, the Enscryb team is incorporating AI into our digital twin software to enable scenario planning and forecasting at scale.
The Bad
Yet, as with any emerging technology, there are pitfalls that require caution.
Liana outlined several key risks:
- Over-Reliance Without Oversight – “We’re a very long way from fully autonomous AI operations,” she cautioned. Critical decisions still require human judgment.
- Bias in Data Models – AI is only as good as the data behind it. Poor quality or incomplete inputs lead to flawed models and bad outcomes - the classic “garbage in, garbage out” problem.
- Cybersecurity Vulnerabilities – AI itself can be exploited. Liana pointed to Google’s recent disclosure that its Gemini model was being manipulated for phishing campaigns - a stark reminder that AI tools can be turned against us if not secured.
- Opaque Governance – Without clear rules on transparency and accountability, AI systems risk becoming “black boxes,” making it difficult for operators and regulators to fully trust their outputs.
The Ugly
Finally, there’s the “ugly” side of AI — its environmental and systemic costs.
- Data Storage – The sheer scale of data required to train and run AI models is “astronomical,” as Liana put it. By 2022, U.S. smart meters alone were producing 54 petabytes of new data annually, consuming massive amounts of energy just to store it.
- Energy Consumption – Training a single model like GPT-3 consumed 1.25 GWh of electricity and emitted 552 tons of CO₂. Add in hyperscale data centers, some of which now require entire new substations to operate, and the strain on the grid becomes clear.
- Regulation & Security – Policymakers are still catching up. Without robust oversight, utilities risk navigating a fragmented regulatory environment that leaves gaps in accountability and protection.
Avoiding the Pitfalls
So how do we avoid the pitfalls? Liana emphasized that the risks are not inevitable - they can be managed with the right strategies:
- Keep humans in the loop – Ensure that AI augments, rather than replaces, human expertise for critical decisions.
- Strengthen data governance – Validate inputs, maintain transparency in model design, and monitor for bias.
- Invest in cybersecurity – Protect AI systems against misuse by embedding security-by-design into tools and processes.
- Be selective with data – Adopt streaming analytics and retention policies that reduce unnecessary storage, lowering both costs and environmental impact.
- Engage with regulators early – Help shape standards and compliance frameworks so AI in the power sector evolves responsibly.
From Insight to Action
Liana’s discussion made clear that while AI offers tremendous promise, the industry must take an intentional and responsible approach.
Liana’s discussion made clear that while AI offers tremendous promise, the industry must take an intentional and responsible approach.
Enscryb’s Digital Toolbox
Our core offering, the Enscryb Digital Toolbox, enables energy stakeholders to simulate, validate, and orchestrate DERs across today’s complex and fragmented value landscape for energy flexibility.
- Enscryb Simulator: A powerful digital twin simulation engine and user-centric platform to model DER systems, run what-if analysis, and right-size investments in a rapidly changing market for operational readiness.
- Enscryb Orchestrator: Optimized predictive control in near real-time across critical assets, distributed energy systems, physical networks, and virtual power plants -- deployable in hybrid environments (Cloud, Edge, or on-premises).
Together, these tools provide the ability to unlock new market models using the same underlying architecture, enable new modes of system operation and aggregation, and deliver both economic and operational optimization.
