Follow

I am working on an analysis of the Commission's leaked legislation plans on Artificial Intelligence. And I would like to hear your opinion on definition. How do you read it? Does that definition fit? Too broad? Too narrow?

· · Web · 4 · 3 · 2

@marcel_kolaja My computer science engineer's opinion is that this definition covers all current forms of software. Which can be deceptive for the public, who generally imagines that AI is some sort of fundamentally distinctive type of software, but I tend to agree with this definition. It's just that I wouldn't bother using the term "AI", software would suffice.

@sossalemaire @marcel_kolaja It would definitely cover all games with their NPC, pluse almost any software which uses at least LRU caching and any OS with non-rrd scheduling.

@marcel_kolaja Všeobjímající "statistical approaches" se už běžně používají při optimalizaci SQL, svým způsobem i JIT/JAVA, dále v operačních systémech a překladačích - kupř. průmysl dodělává do Linuxu podporu pro "safety" (viz aktuální vědecké články o Linux, Safety, Path coverage, Complex systems, atd.) tak za chvíli budou AI nejen hloupé klasifikátory, ale řídící pult ledničky nebo klikomet v mobilu 😀 Bayes mi taky nepřijde moc AI, ale není to svým způsobem jedno, co tam EU napíše? 😩

@g @marcel_kolaja me by zajimala jedna mozna opomijena vec.. totiz jak v pripade OK google tak siri, vyhodnocovani probiha vzdalene.. ten asistent neustale posloucha a neco nekam posila.. je to ci neni to AI? nebo je to jen "browser" , ktery jen zobrazuje "stranku" ? ta vec z EU je hezkej popis, ale spis premyslim nad realnym dopadem a tohle se bude jiste resit, az se ty firmy zacnou vymlouvat, ze to zadna AI neni..

@mig @marcel_kolaja Domnívám se, že to politici hodlají chápat jako AI systém, přičemž je jedno kolik mezičlánků povede do jádra, kde sedí "skutečná AI". Kdyby radši vydefinovali AI jako něco neživého, co má schopnost (samo)učení, tak si ušetří spousty problémů a budou mít kvalitní kladivo proti zneužívání jakékoliv human-killing technologie s vyloučením těch primitivních variant, kde se deterministicky dopředu ví, jestli to někoho zabije nebo ne a není nutné řešit odpovědnost za AI rozhodnutí.

@Marcel Kolaja  AI is more business or popularisation term than any particular mathematical or software technique.

a) is just substitution of AI for "learning"
b) is another substitution for "knowledge"
without definition of either of them

Statistics and Bayes is very funny, but AI is a label for anything. It can be IF ... THEN ... programming clause and the result can be presented as AI.

I think the dangerous algorithms are those which make use of behavioral data and psychological models. Any algorithms which amplify consumption should be banned.

In the Big Hack movie there is a debate about CambridgeAnalytica PsyOps, as a military grade technology. Encryption algorithms once was a restricted software, which could not be exported from US and some maybe still is.

I dont know if algorithm control is possible to define by type of algorithm. A lot of them should be classified. Then it is a stated purpose of using which should be controlled and licenced. It is similar like licence to sell guns or poisonous or explosive chemicals. It should be licenced business.
Sign in to participate in the conversation
Pirátský Mastodon - Sharing is caring

The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!