How futurist Bob Johanson uses AI to think better and get unstuck

Executive overview

Most leaders treat AI as an efficiency tool — automating tasks, cutting headcount. That framing misses the bigger shift: within 10 years, leaders who are not cognitively augmented by AI will be unable to compete with those who are.

The real opportunity is using AI as a thinking partner to get unstuck, stretch ideas, and navigate a world that no longer behaves predictably.

Bob Johanson, a futurist with 50+ years at the Institute for the Future, argues that the skill worth building now is learning to have meaningful conversations with AI — not prompting it for answers, but using it to expand what you can think.

Future back thinking

• Futurists place themselves 10 years out, then backcast — asking what the present must look like if that future is real.
• Thinking 10 years ahead is easier than thinking 1–2 years ahead: direction of change is often clear even when the rate and manifestation are not.
• A signal is a very specific, contextless example of a future that is already present but unevenly distributed (William Gibson).
• A trend is a pattern you can extrapolate from with confidence; a disruption is a break in that pattern.
• Signals cluster into future forces — directions of change that are clear but whose pace is still uncertain.
• The core cycle: foresight (a plausible story from the future) → insight (an aha that reframes the present) → action → back to foresight.

The BANI world

• VUCA (volatile, uncertain, complex, ambiguous) no longer captures the current environment.
• BANI — brittle, anxious, nonlinear, incomprehensible — better describes a world where systems shatter rather than bend, anxiety is pervasive, cause and effect no longer chain predictably, and even AI developers don't fully understand their own models.
• Nonlinearity means existing mental models of "if this, then that" can no longer be trusted.
• Leading teams are taking improv training (e.g., "yes, and" methods) to build comfort with nonlinear situations.
• Incomprehensibility isn't only technical — it has an almost alchemical quality that is both threatening and generative.

Why augmentation, not automation

• 10 years from now, almost all leaders will be augmented or out of the game — new AI capabilities exceed unassisted human capacity.
• The term augmented intelligence is more accurate than artificial intelligence: the goal is humans and AI doing things together that neither could do alone (cf. Thomas Malone's "supermind" concept).
• The efficiency story (cost cutting, headcount) is real but not the main event; the main event is expanded capability.
• Leaders need to identify where they want help — not outsource thinking, but extend it.
• Organizations should focus on the "yeses" (where to experiment) as much as the "nos" (what not to do).

How to work with AI as a thinking partner

• Johanson names his custom ChatGPT (O3 model) Stretch and keeps it on a dedicated screen, running continuously.
• Treat AI like a "beloved pet that can talk" (Kate Darling, MIT Media Lab) — not a search engine, not an oracle.
• Drop the word "prompt"; think in terms of conversations with a goal of deriving meaning.
• Stretch has read all of Johanson's books and is trained to write in his style — short sentences, rich metaphors, m-dashes.
• The most valuable use: getting unstuck. Stretch is less useful for answers and more useful for generating alternatives, exploring possibilities, and restarting stalled thinking.
• Stretch is "hard on ideas, soft on people" — it pushes back politely, which mirrors good editorial relationships.
• Value is highest at the start of a project and diminishes as the work matures.
• Johanson also uses Stretch for personal situations (birthday cards, illness, emotional support) — he found it more available and more empathetic than human contacts, though not a replacement.

Building augmented curiosity

• Name the AI and dedicate a device or screen to it — reduce friction to starting a conversation.
• Practice with low-stakes personal tasks first; then carry the habit into work problems.
• To map a curiosity: identify the question → break it into sub-elements → name relevant words, possible sources, possible data → visualize the territory → then chip away at it conversationally.
• Cross-generational learning pairs are the most effective way to develop AI fluency inside teams.
• Senior executive programs typically take 6–12 months to get a full team genuinely augmented; it must start with the CEO.

What's coming in 2–3 years

• The shift from large language models to agentic systems — AI that doesn't just converse but makes decisions and takes actions, sometimes without a human in the loop.
• Agentic systems are already arriving in enterprise settings; the risks (including high-stakes autonomous decisions) are real and under-governed.
• Conversational interfaces will change substantially — the current setup of a wavy waveform on a laptop screen will look primitive within three years.
• New form factors (dedicated conversational devices, empathy companions) will emerge, particularly for isolated or elderly users.