When DeepDream for audio is invented, the output will cause seriously ill effects in humans like seizures. This will happen before 2020.
Created by Houshalter on 2016-02-09; known on 2020-01-01; judged wrong by gwern on 2020-01-01.
- Houshalter estimated 25% on 2016-02-09
- gwern estimated 10% and said “I don’t recall any instances of vomiting or epilepsy in reaction to DeepDream images…” on 2016-02-09
- Houshalter said “Right, my theory is that audio is fundamentally different. Requiring recurrency and having less variance with scale/position.” on 2016-02-09
- molochhasme estimated 1% and said “Not likely” on 2016-02-09
- JoshuaZ estimated 2% and said “It is much easier to cause ill effects in humans now through the visual medium than auditory medium (e.g. epilepsy). ” on 2016-02-10
- Raahul_Kumar estimated 0% and said “This would require an audio version of Deepdream to be invented first, and then for it to have ill effects like seizures.” on 2016-02-10
- themusicgod1 estimated 4% and said “https://www.reddit.com/r/experimentstodo/comments/4562vz/audio_deepdream/” on 2016-02-11
- PseudonymousUser estimated 10% on 2016-02-11
- PseudonymousUser estimated 5% on 2016-02-14
- sole21000 estimated 2% and said “Assuming it’s limited to the regular musical sound range, I don’t really see what mechanism would effect it.” on 2016-03-09
- trishume estimated 1% on 2016-05-09
- pranomostro estimated 1% on 2018-09-02
- AlexLamson estimated 0% and said “I see no reason why such ill effects would happen.” on 2019-05-30
- PseudonymousUser estimated 0% on 2019-09-04
- gwern judged this prediction wrong on 2020-01-01.