What Happened?
Nowadays, a growing number of people place unconditional trust in generative AI — a phenomenon that spans both the input and output sides of how humans engage with knowledge.
Some argue that AI has drastically boosted human productivity and accelerated the speed at which people grasp knowledge. It integrates seamlessly and effortlessly into processes ranging from requirement definition to validation in product development, and from conceptual comprehension to mastery in learning, acting as a capable colleague or an approachable tutor throughout.
Yet when we step back and examine the vast majority of outputs fully or largely driven by AI, one easily overlooked yet critical point remains to be verified: whether they truly address our actual needs without introducing extraneous pseudo-needs irrelevant to the core objective — that is, whether they qualify as "demand-pure" deliverables.
For sure that this concern implicitly overlooks the precision of user inputs. Most of the time, people do not invest extensive effort in meticulously crafting prompts to obtain a perfect result from AI. In many cases, it's unavoidable that user queries already contain a string of imprecise phrasing or logically muddled instructions.
This naturally raises a further question: as we carry on continuous dialogue with AI, how can we ensure our original demands remain constant, and not shaped or altered by the AI’s outputs? If our cognition is shifted by new concepts introduced by the AI during this process, it becomes extremely difficult to guarantee that the needs we held in the initial state will remain unchanged.
My Reflections
- Premise 1: The cognitive gap between users and AI cannot be eliminated.
- Premise 2: Users cannot ensure their expressions perfectly and accurately align with their actual needs when using AI tools.
- Premise 3: By their very nature, AI outputs are non-deterministic products generated through probabilistic inference.
- Premise 4: Users cannot guarantee that their needs remain logically closed and constant over the course of interacting with AI.