an agreement you cant read is still an agreement thatโs the problem
(this example is hypothetical) picture two people signing the same terms one reads them in their first language the other reads them in a language they use at work but not at home and skims past @GenLayer
one model can be confidently wrong ask it whether a source is recent enough and it may answer yes just as firmly whether or not itโs right @GenLayer validators run different ai models and each one assesses the outcome on its own no single answer decides
@GenLayer with it
โ a rule nobody can read is a rule only the writer can use
โ a rule thatโs clear to one side and vague to the other is a quiet advantage
โ a result thatโs explained in plain words makes the terms easier to check
@GenLayer it compares them under its rule
agreement isnโt proof the models only see the evidence theyโre given so weak evidence gives weak agreement still one output with no second opinion is a thinner base to stand on
one model can be confidently wrong ask it whether a source is recent enough and it may answer yes just as firmly whether or not itโs right @GenLayer validators run different ai models and each one assesses the outcome on its own no single answer decides
the time between a token arriving and a network going live is easy to overlook itโs also where many people decide whether theyโll take part at all @GenLayer
GenLayerโs pre mainnet staking is built around that gap. as announced on 2 november from THE DAY unclaimed airdrop GEN
@GenLayer the result validators donโt need to phrase their answers the same way only to reach an equivalent conclusion under the rule the contract sets
thatโs the point of using different models when they answer differently the contract doesnโt have to trust whichever one sounded surest