Erik Witte

Is AI a rising superintelligence or just a "normal technology"? In an essay for the Knight First Amendment Institute, Prof. Arvind Narayanan and Sayash Kapoor argue that treating AI as an autonomous, superintelligent entity misinterprets its true societal trajectory. Instead, framing AI as "normal technology"—akin to electricity or the internet—reveals a far more realistic future. The essay is written in 2025 and technology has evolved since then so here's a link to the latest work - https://lnkd.in/dP7uMhC6 as well as 2025 essay - https://lnkd.in/dwsY2Cr3. Here are three key insights from their work: • Diffusion is slow: While AI capability benchmarks advance rapidly, real-world adoption in high-consequence sectors unfolds over decades due to safety, regulatory, and organizational constraints. • Humans remain central: Future workplace roles will center on AI control, monitoring, and task specification rather than total human replacement. • Prioritise resilience over panic: Rather than focusing heavily on speculative existential threats, policy should target concrete systemic risks by building societal resilience. Viewing AI as a powerful tool rather than an uncontrollable force allows us to design better safety controls, system audits, and governance frameworks. https://lnkd.in/dwsY2Cr3

15

The go forward risk in 2026 is not implementing AI poorly; it’s ignoring it while your competitors use it to diligence deals faster and run PortCos cheaper. As #EQT philosophy suggests: AI won't replace investment firms, but investment firms that use AI will replace those that don't. Juhi Saxena Rachel Barton Vivek Narayanan

19

2 Comments

AI is becoming a commodity. Models are accessible. Vendors are comparable. Tools are everywhere. So where does competitive advantage come from now? I just read a powerful piece in Harvard Business Review by Rohan Narayana Murty and Ravi Kumar S that reframes the debate: 👉 When every company can use the same AI models, context becomes the differentiator. Not documentation. Not CRM stages. Not SOPs. But the invisible patterns that define how work actually gets done: • When do leaders escalate? • How much uncertainty is tolerated? • Which signals predict deal slippage? • When is speed prioritized over certainty? • What trade-offs define “how we win”? This is institutional memory. And here’s the shift: AI models are general by design. Your organization’s operating logic is not. Without contextual grounding, AI standardizes behavior. With contextual grounding, AI amplifies competitive advantage. The next frontier is not “AI deployment.” It is Context Engineering: → Capturing execution patterns → Encoding tacit judgment → Delivering precise context at decision time → Linking AI actions to business outcomes In the next 2–3 years, we won’t compete on model access. We will compete on how well we encode how we win. For enterprises — especially in BFSI, regulated sectors, and IT services — this may be the real AI moat. The question for CXOs is no longer: “Which model are we using?” It is: “Have we captured our operating logic?” #AI #GenerativeAI #Leadership #DigitalTransformation #EnterpriseAI #Strategy #CXO #Innovation

5