AI with Accountability: Why Validation Matters More Than Hype
Jon Brewton, CEO, Data2, stops by, and we talk about accountability.The negative issues we are facing about Data Centers and AI are hitting the front pages, and we need accountability for AI. We also need accountability for local politicians and all elected officials.Jon just presented right behind the Head of AI in Energy for Microsoft in an AI presentation. There were several key points in the Microsoft presentation that there was no company that could scale accountability across disparate systems. After Jon presented, the Microsoft presenter came up and said you may be the only company in the world that can provide accountability across disparate systems at scale. This is huge.When the AI lead presenter comments, “Your presentation just dispelled everything that I said that there is no accountability in AI.” This is critical.1. AI in the Energy IndustryThe conversation centers on how artificial intelligence is transforming the energy sector. While AI is positioned as a major strategic opportunity, there’s a critical distinction between AI being impressive and AI creating real, measurable business value. The speakers emphasize that many companies are investing heavily in AI without seeing corresponding productivity gains.2. The Hallucination Problem & AI ValidationJohn Bruton discusses Data Squared’s patented solution for addressing AI “hallucinations”—instances where AI systems generate plausible-sounding but incorrect information. The core issue is that AI without validation and cross-checking is worthless. The patent focuses on creating explainable, trustworthy, and auditable AI results, allowing organizations to “lift the hood” and see exactly how the system is reasoning.3. Data Integration & Legacy SystemsA major challenge in the energy industry is connecting disparate, fragmented systems that have accumulated over decades. The speakers highlight a real-world example where an oil company with eight acquired subsidiaries had billing processes taking 2 months; using Data Squared’s solution, they reduced it to 2 minutes with verification. The key insight: you cannot create sustainable AI value without solving structural data integration problems first.4. The Four Major AI MisconceptionsLLMs aren’t the solution: Large language models (ChatGPT, Gemini, etc.) are only a small component; the real challenge is data orchestration, context continuity, and workflow integrationMore data ≠ better AI: Simply aggregating all available data doesn’t improve AI; inconsistent definitions and siloed information actually create problemsInfrastructure forecasting is based on promise, not proven demand: Data center expansion is being driven by theoretical AI adoption, not current production-grade deploymentsAI should augment humans, not replace them: The goal should be human-machine collaboration, not automation that eliminates jobs5. Data Center Infrastructure & Eminent DomainThe conversation addresses growing public concern about large data centers being built on private land. The speakers propose an alternative: small-scale, distributed data centers connected via mesh networks, placed next to existing energy sources (stranded gas, geothermal, etc.). This approach could:Reduce infrastructure footprint by ~90%Cut upfront capital costs by ~90%Reduce energy consumption by ~90%De-risk infrastructure at scale6. Federal Government & Legacy SystemsData Squared is working with federal agencies (DoD, VA, FBI) to integrate decades-old systems. The VA alone has operated systems from 1947, 1956, 1963, and the 1980s—all disconnected. The speakers emphasize that government is “probably the worst offender” for unintegrated legacy systems.7. Nuclear Energy & AI IntegrationThe speakers discuss the growing role of nuclear power (including small modular reactors) in supporting AI infrastructure. They emphasize that nuclear applications require absolute reliability, transparency, and auditability—making Data Squared’s approach particularly valuable.8. ROI & Business ValueA recurring theme is the CEO question: “I’m spending millions on AI—where’s my return?” The speakers argue that with proper validation and integration, AI can deliver measurable bottom-line value, but only when the structural data problems are solved first.This is huge, with accountability, AI can be rolled out, and we need accountability across AI and the political landscape.As I mentioned in my last SubStack article, we need political accountability at the national, state, and local levels. Get involved and help provide accountability at the local levels, and keep us posted.We need data centers rolled out responsibly and not hurting consumers. With accountability, this can be done.We have several more topics and interviews with Jon and other CEOs in the AI and Energy space on the drawing board, covering nuclear, natural gas, midstream, downstream, oilfield services, exploration, and especially utilities.Connect with Jon on his LinkedIn, and tell him Stu sent you. https://www.linkedin.com/in/jon-brewton-datasquared/Data2 if you have any business systems, can you trust A? Well, they have the patent on validation. https://data2.zoholandingpage.com/energySubstack article https://theenergynewsbeat.substack.com/p/ai-with-accountability-why-validation
