Generative AI at Work / Unit 2 GAI 047 worked example, 60 to 90 seconds
Trade-offs Between the Marks
The idea
NIST's framework is explicit that its seven characteristics cannot all be maximized together: improving one often costs another, and the balance depends on how the system is used.
In practice
Step one: name the use, because NIST says the right balance depends on the context of use. Step two: look for the known tensions. NIST's examples are interpretability against privacy, predictive accuracy against interpretability, and privacy protections that cost accuracy when data is sparse. Step three: decide which characteristics matter most here, and record why. NIST asks that trade-offs be resolved in a way that is transparent and justifiable. Step four: check the weakest characteristic, because NIST describes trustworthiness as only as strong as its weakest part.
Where people slip
Strength on one characteristic never makes up for failure on another; NIST's own example is that a secure but unfair system is still undesirable.
Your place on the trail
Every lesson rests on earlier ones and opens later ones. Tap any to jump.
Unlocks
- Nothing yet depends on this.
Something wrong in this lesson? Flag it · Sources for Generative AI at Work