Engineered Intelligence and The Data Science Problem in AI

Questioning replaces coding as literacy.
AI has shifted the fundamental definition of data literacy from technical execution (like writing Pandas code) to "contextual application." The core skill for the modern worker is the ability to ask the right questions and interpret probabilistic results rather than mastering specific software syntax.

Human data misuse dwarfs AI hallucinations.
Leaders frequently over-index on the risks of machine error while ignoring the fact that humans have "hallucinated" with data for decades. The primary threat isn't a probabilistic model being wrong; it's the lack of human critical thinking and emotional intelligence to verify and apply the model's output correctly.

Organizations are repeating 2012's mistakes.
Just as companies failed to extract value from "the sexiest job of the 21st century" by hiring data scientists without building a data-literate culture, they are currently buying AI tools while neglecting the fundamental change management and mindset shifts required for the 99% of non-technical staff to use them.

IQ and EQ anchor Engineered Intelligence.
Jordan Morrow defines the formula for successful decision-making as "Engineered Intelligence": two parts data and AI combined with two parts human IQ and EQ. This framework ensures that AI augments rather than replaces the human intuition, gut feel, and empathy necessary for high-stakes business outcomes.

Shadow AI proves individual value.
While 90% of enterprise-level generative AI projects are reportedly failing, "Shadow AI"—where individuals use tools outside of official channels—is thriving. This disconnect suggests that the value is real and immediate at the task level, but enterprise governance and product wrappers are lagging behind individual curiosity.

AI identifies human blind spots.
Rather than just making old processes more efficient, leaders should use AI as a "PhD partner" to identify what humans are missing. By feeding deep research into models and asking for the "hidden pieces," teams can overcome the structural biases of their own mental models to prepare for Black Swan events.

AI ROI demands a two-year horizon.
True return on investment from AI is typically 12 to 24 months out, but many leaders are too "dopamine-driven" to wait. Success requires a willingness to accept short-term losses and dedicated "experimentation time" for employees to ensure long-term competitive gains.

Data stacks fail without networking.
The most successful Chief AI Officers don't win by building the perfect technical architecture in isolation. They succeed by "leaving their desks" and acting as change management specialists who bridge the gap between technical capability and on-the-ground business problems through constant internal networking.

Predictive modeling is the overlooked quick win.
While leaders chase generative AI, mature ML-based predictive modeling has been delivering reliable business value for decades. Jordan steers clients toward predictive use cases first: take the data you already have, build predictions that "see around the corner," and capture ROI before investing in more complex generative or agentic systems.

Six of the top ten fastest-growing skills are human.
The World Economic Forum projects that by 2030, adaptability, creativity, analytical thinking, and similar human capabilities will dominate the fastest-growing skills list. As AI automates more tasks, the competitive edge shifts to people who can problem-solve, communicate, and think critically, not those who know the most tools.

You can read the full transcript here.