The AI Mode Choice
Every AI session opens in one of two operating modes; the one chosen in the first thirty seconds shapes the output more than any prompt that follows.
The Mode Choice is the framework for deciding, before any prompt is typed, whether to lead with your own context and judgement or to lead with AI’s suggestions. This was named and refined through ChangeSchool’s work with senior leaders across our executive education programmes, as part of the modular blocks of our The Art of Delegation curriculum.
​
The framework: Input-First and Suggestion-First
Two ways to open any AI session, each with its own strengths and its own failure mode.
​
The Mode Choice, a fork between input-first (context-rich, frame-stable; failure mode: polished version of a wrong frame) and suggestion-first (frame-uncertain; failure mode: the first AI list becomes the menu). Mixed mode is a deliberate two-step sequence — suggestion-first, pause, input-first.
Input-first. The leader gives AI their context, constraints, history and judgement before asking for anything generative. The things AI cannot know on its own go in first: what the team can carry, what has been tried, the political reality the decision sits inside, the half-formed view already taking shape. AI is then asked to develop, critique or extend that thinking. The output carries the leader’s fingerprints. The trade-off is narrowness; if the frame the leader brings is wrong, input-first produces a polished version of a wrong frame.
​
Suggestion-first. The leader asks AI for options, framings or approaches before anchoring it with context. The prompt is lean, the context is held back, and AI reaches into its training to offer angles the leader had not considered. The output widens the search space. The trade-off is that the first list AI produces becomes the menu from which the leader selects, and judgement arrives late, after the anchor has already taken hold.
​
Two different moves, with two different outputs and two different failure modes when each is used in the wrong situation.
​
The decision rule
Context-rich, frame-stable: input-first. Substantial leader context that AI does not have, and reasonable confidence the framing is right. Strategy memos for a known business; decisions where constraints are specific; communications where relationships and history matter.
​
Frame-uncertain: suggestion-first. Doubt about whether the current frame is the right one, or a sense that being inside the situation has narrowed the view. New-market questions, strategic re-thinks, anything where the honest answer to am I thinking about this the right way is I am not sure.
Reframing then execution: mixed mode, in that order. Most consequential work needs both, with suggestion-first first. Once context is loaded, the framings AI offers become variations on the leader’s frame rather than alternatives to it. The order is not interchangeable: suggestion-first must run before input-first, or the wider search space never opens.
​
Genesis
Two older findings sit underneath. Daniel Kahneman and Amos Tversky (1974) established that whatever value or framing is presented first pulls subsequent thinking towards it, with adjustment rarely sufficient to overcome the pull. Evan Risko and Sam Gilbert (2016) established that when a tool can do the thinking, people let it, often without noticing; they called this cognitive offloading. AI is the most capable cognitive offloading tool yet built, and whatever it produces first becomes the anchor for the leader’s reasoning. The two effects compound.
​
Why it matters now
Four forces pull leaders to the wrong opening mode. The cost gradient runs one way: suggestion-first is cheap, input-first requires ten minutes of writing before any work begins. Time pressure favours the cheap mode on every task. Confidence calibration misleads in both directions; insecure leaders reach for suggestion-first to be told what to think, confident leaders reach for input-first to be confirmed in what they already think. And the reflex built by hundreds of small frictionless AI interactions (the Candy Machine Trap, virenlall.com/candy-machine-trap) becomes the reflex applied to consequential work.
​
The moves
Three corrective moves, each takes thirty seconds or less.
​
Name the mode before writing the prompt. Ask one question of yourself: do I have context AI does not, and is my frame stable; or is my frame uncertain; or both, in sequence? The answer names the opening mode for this session.
​
Hold context back when suggestion-first is right. The leanest possible problem statement, no preferred direction, no for my situation or given that we are. Each addition narrows the search space towards the frame already being held; the point of suggestion-first is to widen it.
​
Switch modes deliberately, with a pause between. Mixed mode is two interactions, not one continuous flow. Run suggestion-first, read the outputs, step away, choose the framing with reasons that can be articulated, then start a new session for input-first execution. The pause is where the leader’s judgement re-enters before the second mode begins.
​
The discipline
The corrective moves work on a single task. Three habits embed the practice.
​
A thirty-second pause built into the prompting reflex, every consequential session. The habit is the pause itself; the answer varies by task.
Mode-tagging of AI sessions in whatever tool tracks the work. Note input-first, suggestion-first, or mixed at the top of each session. After a month, the pattern in the defaults becomes visible; most leaders find they have been tagging eighty per cent of sessions as suggestion-first, on tasks that needed otherwise.
Periodic mode-swapping on completed work. Once a fortnight, take one piece of AI-assisted work done in the default mode and run it again in the other mode. The two outputs sit side by side, and what the default has been hiding becomes visible.
​
The discipline is to make the Mode Choice deliberate every time, and to keep checking the defaults that were never noticed in the first place. The pattern across a quarter, captured by the tags, becomes the input the leader uses to retrain their own first reflex.
​
​
​
​
‘The mode you start in is the mode the output remembers.’
Viren Lall, Managing Director,
ChangeSchool LDN (2026).
virenlall.com/ai-mode-choice
​
​
​
AI for Leaders.
Executive Education that changes practice.
Viren Lall is Managing Director of ChangeSchool LDN, a London-based executive education partner. ChangeSchool specialises in AI for senior-leader development, winning the EFMD Global Excellence in Practice Award in 2023 and 2025, with programmes in 39 countries.
​
Since April 2024, ChangeSchool LDN has been designing and delivering mindset shifts through Executive Education Programmes across sectors such as deep tech, manufacturing, and education, for business owners, governance professionals, and senior leaders. Leaders gain AI fluency, protect decision quality, spot value creation opportunities, and foster human-centric AI use. AI capability for senior leaders is also a core element and a constant spine of our Open Programmes for Chief Digital Officers, Chief Operating Officers, and Chief People Officers, delivered by our partner business schools.
Some of our clients include the Royal Academy of Engineering, Education and Training Foundation, and the UK Government's Meet Smart programme.
For speaking, programme, or partnership enquiries, get in touch with him through ChangeSchool LDN.