Viewing a response to: @pedrobrito2004/catfishaiandtherapeuticswhatdotheyhavetodowitheachother-sj9tbwkrm6
There is a very simple hack for computers... or AI that humans understand and for them, it's still hard... which is... # Giving up... Usually codes are made to iterate around things... and they necessarily give up for "natural" causes (usually if they do, it's because they found you overtaking the rules of engagement, like being negative or verbally rude, etc). So, in some simple techniques it's quite easy to test and FAIL AI algos... like, for example when you repeat the same answer several times over and over but using different words... or even repeating 100 times the same exact sentence (unless they block you for repetition reasons). The other simple one, is when generate a normal response, and then contradict your self, then repeat... The most common problem with the above is emotions and disposition, which is already somehow trialed on AI, but way imperfect to match humans. So when we give up due a "made up" reason, sometimes the AI can't understand why. If you do it several times and never change... they will randomly fail more often.
| post_id | 80,080,013 | ||||||
|---|---|---|---|---|---|---|---|
| author | forykw | ||||||
| permlink | re-pedrobrito2004-2019917t215734642z | ||||||
| category | steempress | ||||||
| json_metadata | {"tags":["steempress","blog","pedrobrito2004"],"app":"esteem\/2.2.0-surfer","format":"markdown+html","community":"esteem.app"} | ||||||
| created | 2019-09-17 09:57:36 | ||||||
| last_update | 2019-09-17 09:57:36 | ||||||
| depth | 1 | ||||||
| children | 2 | ||||||
| net_rshares | 9,521,070,700 | ||||||
| last_payout | 2019-09-24 09:57:36 | ||||||
| cashout_time | 1969-12-31 23:59:59 | ||||||
| total_payout_value | 0.000 SBD | ||||||
| curator_payout_value | 0.000 SBD | ||||||
| pending_payout_value | 0.000 SBD | ||||||
| promoted | 0.000 SBD | ||||||
| body_length | 1,061 | ||||||
| author_reputation | 24,990,662,983,063 | ||||||
| root_title | "Catfish, AI and Therapeutics, what do they have to do with each other?" | ||||||
| beneficiaries |
| ||||||
| max_accepted_payout | 1,000,000.000 SBD | ||||||
| percent_steem_dollars | 10,000 | ||||||
| author_curate_reward | "" |
| voter | weight | wgt% | rshares | pct | time |
|---|---|---|---|---|---|
| crypto.piotr | 0 | 9,159,858,359 | 1% | ||
| pedrobrito2004 | 0 | 361,212,341 | 100% |
Thanks for the comment. Certainly, the actuality of AIs fall into notorious failures when facing things like that. I have the idea that in the future, the introduction of neural networks of greater capacity and the use of memory meshes that mimic those possessed by mammals can help overcome this. Maybe the future solution will take a different path than I imagine, but it will surely be something interesting.
| post_id | 80,082,357 |
|---|---|
| author | pedrobrito2004 |
| permlink | pxz5fm |
| category | steempress |
| json_metadata | {"tags":["steempress"],"app":"steemit\/0.1"} |
| created | 2019-09-17 12:02:24 |
| last_update | 2019-09-17 12:02:24 |
| depth | 2 |
| children | 1 |
| net_rshares | 9,156,374,008 |
| last_payout | 2019-09-24 12:02:24 |
| cashout_time | 1969-12-31 23:59:59 |
| total_payout_value | 0.000 SBD |
| curator_payout_value | 0.000 SBD |
| pending_payout_value | 0.000 SBD |
| promoted | 0.000 SBD |
| body_length | 413 |
| author_reputation | 49,990,662,111,101 |
| root_title | "Catfish, AI and Therapeutics, what do they have to do with each other?" |
| beneficiaries | [] |
| max_accepted_payout | 1,000,000.000 SBD |
| percent_steem_dollars | 10,000 |
| author_curate_reward | "" |
| voter | weight | wgt% | rshares | pct | time |
|---|---|---|---|---|---|
| crypto.piotr | 0 | 9,156,374,008 | 1% |
Yes... but that proportions (right now) are still under HPC (High Performance Computing) arena, where you need a large amount of resources to really do "WOW" things. I have some experience with Genetic Algos under neural networks... and yes I agree with some extent with you... but I think selective intelligence on neural networks (to be more specific) is where things will evolve. Big meshes... will always cost a lot to process... unless they decide a way to monetize this very well on things like EOS... which was what I though things would go naively when I saw C++ as programming language for contracts as a possibility. The blockchain world still has to meet large scale concentrated resources (which is not yet a desirable exploratory area), like MPI (message passing interface) language/libs... but I am certain that in some cases this will empower companies a LOT! Depends on the applications... but specially if we combine neural networks or AI in general with Blockchain... damn... it's going to get crazy on my side... and maybe then, my employer will hide me in a cave... LOL =) when they know what I know about blockchain...
| post_id | 80,082,556 | ||||||
|---|---|---|---|---|---|---|---|
| author | forykw | ||||||
| permlink | re-pedrobrito2004-2019918t01125950z | ||||||
| category | steempress | ||||||
| json_metadata | {"tags":["steempress"],"app":"esteem\/2.2.0-surfer","format":"markdown+html","community":"esteem.app"} | ||||||
| created | 2019-09-17 12:11:27 | ||||||
| last_update | 2019-09-17 12:11:27 | ||||||
| depth | 3 | ||||||
| children | 0 | ||||||
| net_rshares | 9,538,695,273 | ||||||
| last_payout | 2019-09-24 12:11:27 | ||||||
| cashout_time | 1969-12-31 23:59:59 | ||||||
| total_payout_value | 0.000 SBD | ||||||
| curator_payout_value | 0.000 SBD | ||||||
| pending_payout_value | 0.000 SBD | ||||||
| promoted | 0.000 SBD | ||||||
| body_length | 1,141 | ||||||
| author_reputation | 24,990,662,983,063 | ||||||
| root_title | "Catfish, AI and Therapeutics, what do they have to do with each other?" | ||||||
| beneficiaries |
| ||||||
| max_accepted_payout | 1,000,000.000 SBD | ||||||
| percent_steem_dollars | 10,000 | ||||||
| author_curate_reward | "" |
| voter | weight | wgt% | rshares | pct | time |
|---|---|---|---|---|---|
| crypto.piotr | 0 | 9,154,599,146 | 1% | ||
| pedrobrito2004 | 0 | 384,096,127 | 100% |