ruti on Nostr: yes, its important that companies working on ai are using unbiased data sets. i ...
yes, its important that companies working on ai are using unbiased data sets. i assume alot of data labelling is done manually by humans on what is an ‘appropriate’ response and even the humans doing this can be biased. i agree with you 100% on that point.
but if an AI has enough data and enough nodes/parameters, it will be able to output correct results- everytime. thats why i say it’s only going to get better the more training a model receives. the more eyes developing AI, the less biased it will be. ive already noticed chatgpt is a quite liberal, and i *do* have a problem with that. ai shouldnt be biased in either way. if we achieve general ai or the singularity - we bulld an ai that accurately reflects all of human knowledge, it wouldnt be catastrophic, i dont think. it could revolutionize and optimize learning, get rid a lot of boring jobs (ubi is inevitable), and provide helpful advice to people that need it. it can be your therapist as well as your teacher. i think *thats* where we’re headed, and im not really opposed to it. the good heavily outweighs the bad imo
but if an AI has enough data and enough nodes/parameters, it will be able to output correct results- everytime. thats why i say it’s only going to get better the more training a model receives. the more eyes developing AI, the less biased it will be. ive already noticed chatgpt is a quite liberal, and i *do* have a problem with that. ai shouldnt be biased in either way. if we achieve general ai or the singularity - we bulld an ai that accurately reflects all of human knowledge, it wouldnt be catastrophic, i dont think. it could revolutionize and optimize learning, get rid a lot of boring jobs (ubi is inevitable), and provide helpful advice to people that need it. it can be your therapist as well as your teacher. i think *thats* where we’re headed, and im not really opposed to it. the good heavily outweighs the bad imo