AI will pitch a show for your conference, but it will not pick the right one

The main danger of AI in creative tasks is not bad ideas but too many of them. Ask it to come up with the opening of a conference and you get light, music, video, interactive elements and shows in dozens of combinations.

At the first stage that is convenient: the blank page disappears, you can quickly assemble a wide set of directions and ask the model to make the options bolder. But the number of options does not by itself bring you closer to a good decision.

Because the opening of a conference is not a special-effects contest. It has to land with a particular audience, the mood of the event and the expectations of the people in the hall. An idea that looks striking in a description can turn out inappropriate at the actual event, or simply get lost among familiar devices.

So AI here is more useful as a generator of raw material than as a director. The model helps widen the search field, but the criteria for choosing still have to be set by a human: what we want to say, to whom, and why this particular format will work.

There is an important limitation in the source material too: it shows only the opening of the story and does not reveal which decision finally worked. So no conclusion about a specific opening format can be drawn. A more general one can: the more ideas AI produces, the more important it is to define in advance the frame by which the surplus will be discarded.

When did you last get from AI not too few ideas but too many — and by what criterion did you pick one? Source: All articles / Artificial Intelligence / Habr