This is why we need more experiments like the one in Finland. It would be irresponsible to implement a UBI until we know which of these two scenarios is most likely. Although, I suspect the reality will fall somewhere between the two.

Or, just implement it and adjust the payment for stability.

There's not any data to support it because there's no data around a nationwide, long-term UBI, period. All we have are little experiments where people know it could end at any time.

Qualify that enough, you can make the point. But UBI has been around long enough to collect data (pensions, Alaska, experiments). What people 'know' has no place in economics. We can measure - did prices rise in those places? No? Yes? How much?