[rubendv]

I guess it is in OpenAI's best interest to downplay the memorization aspect in favor of the logical reasoning angle. If it turns out that GPT is memorizing and reproducing copyrighted data, it could land them in legal trouble.

[brookst]

On the bright side it would mean they have invented an amazing compression algorithm, given the model size and amount of text it can produce.

[dragonwriter]

> I guess it is in OpenAI’s best interest to downplay the memorization aspect in favor of the logical reasoning angle. If it turns out that GPT is memorizing and reproducing copyrighted data, it could land them in legal trouble.

Its also in their interest, since it encourages people to attempt to build businesses on top of its “logical reasoning” capacities.

And as long as its within the realm of spin rather than direct false statements, it probably avoids creating them legal problems for fraud, although the difference in effect is…nonexistent.