8,683
edits
Talk:Restless Souls/Technology: Difference between revisions
Talk:Restless Souls/Technology (view source)
Revision as of 21:26, 10 February 2026
, Yesterday at 21:26no edit summary
Paradox-01 (talk | contribs) mNo edit summary |
Paradox-01 (talk | contribs) mNo edit summary |
||
| Line 555: | Line 555: | ||
The questions are too often binary - is it or is it not - and therefore we are tempted to give binary answers. To think that way was a mistake. We would have more accurate and fair discussions if we respect the nature of this subject. | The questions are too often binary - is it or is it not - and therefore we are tempted to give binary answers. To think that way was a mistake. We would have more accurate and fair discussions if we respect the nature of this subject. | ||
Do you remember discussions about the intelligence quotient? Its existence shows that also human intelligence is not binary, possessing it or possessing it not. In fact there is even criticism about it for just being one number expressing an | Do you remember discussions about the intelligence quotient? Its existence shows that also human intelligence is not binary, possessing it or possessing it not. In fact there is even criticism about it for just being one number expressing an average. There are multiple (sub) forms of human intelligence. | ||
* '''ANI''' = Artificial Narrow Intelligence, '''specialized or "weak AI"''' (basically machine learning: pattern recognition) | * '''ANI''' = Artificial Narrow Intelligence, '''specialized or "weak AI"''' (basically machine learning: pattern recognition) | ||
| Line 568: | Line 566: | ||
:: After the obvious slowdown in advancement through scaling, chain of thought (COT) was introduced. It is also known under the marketing term "reasoning [AI]". (See: Gemini 2.0 Flash Thinking and ChatGPT o3, the later one especially [https://techxplore.com/news/2024-12-ai-human-general-intelligence.html trained to cheat … erm ... score high in the ARC-AGI test].) | :: After the obvious slowdown in advancement through scaling, chain of thought (COT) was introduced. It is also known under the marketing term "reasoning [AI]". (See: Gemini 2.0 Flash Thinking and ChatGPT o3, the later one especially [https://techxplore.com/news/2024-12-ai-human-general-intelligence.html trained to cheat … erm ... score high in the ARC-AGI test].) | ||
:: This pushes ANI somewhat more into direction of AGI with the drawback of slower response times and higher energy costs.--> | :: This pushes ANI somewhat more into direction of AGI with the drawback of slower response times and higher energy costs.--> | ||
* '''Physical AI''' = Physical Artificial Intelligence. Basically AI used in robots. Since direct training would be dangerous, slow and therefore ineffective, the AI gets pre-trained in a simulation where the robot is represented by a digital twin. By the nature of this construction, multimodal learning (MML) is predestined for robots. Like humans (or other real organisms), AIs should include an "inner world" for better understanding. The use of LLM is optional but seems to be a good design choice for walking further into direction of AGI. <!--Expecting [[wp:Symbolic_artificial_intelligence#Neuro-symbolic_AI:_integrating_neural_and_symbolic_approaches|some kind of hybrid approaches]] within physical AIs later when all other current ''cash cow'' approaches reached their end.--> | |||
* '''AGI''' = Artificial General Intelligence, also "strong AI" ('''on par with human thinking''', ''a real AI'' capable to fully self-improve and drive its own development) | * '''AGI''' = Artificial General Intelligence, also "strong AI" ('''on par with human thinking''', ''a real AI'' capable to fully self-improve and drive its own development) | ||
:: In discussions AGI is often equated with super intelligence. The argument is that as soon as AGI is achieved ASI is just around the corner. | :: In discussions AGI is often equated with super intelligence. The argument is that as soon as AGI is achieved ASI is just around the corner. | ||
| Line 841: | Line 840: | ||
'''Anthropomorphisierung III'''<!--Schwerpunk: Die Illusion von Interaktion. Oder sollte man es eine Hysterie nennen?--> | '''Anthropomorphisierung III'''<!--Schwerpunk: Die Illusion von Interaktion. Oder sollte man es eine Hysterie nennen?--> | ||
Das stupide (fehleranfällige und im negativen Sinne automatische, nicht-intelligente) Kopieren von Muster durch Bots alleine ist schon kritikwürdig, aber der '''[[wp:Moltbook|Moltbook]]-''Hype''''' hat dem Ganzen | Das stupide (fehleranfällige und im negativen Sinne automatische, nicht-intelligente) Kopieren von Muster durch Bots alleine ist schon kritikwürdig, aber der '''[[wp:Moltbook|Moltbook]]-''Hype''''' hat dem Ganzen die Krone aufgesetzt. | ||
[https://www.zdfheute.de/wissen/ki-agenten-netzwerk-moltbook-100.html zdfheute.de (8. Februar)]: "Social Media für KI: Entwickeln Bots ein Eigenleben?" | [https://www.zdfheute.de/wissen/ki-agenten-netzwerk-moltbook-100.html zdfheute.de (8. Februar)]: "Social Media für KI: Entwickeln Bots ein Eigenleben?" | ||
* Oberer Videobeitrag (Videodateiname: "Neues Netzwerk Moltbook Entwickeln KI-Bots ein Eigenleben") Bei Sekunde 37 ist die Rede von Reaktionen durch viele weitere Bots. Das soll exemplarisch für | * Oberer Videobeitrag (Videodateiname: "Neues Netzwerk Moltbook Entwickeln KI-Bots ein Eigenleben") Bei Sekunde 37 ist die Rede von Reaktionen durch viele weitere Bots. Das soll exemplarisch für das Netzwerk stehen. Es stellt sich raus, viele (auch Leitmedien) wollten etwas in Moltbook sehen, dass da so gar nicht drin steckt.<!-- | ||
31. Januar | 31. Januar | ||
* https://www.heise.de/news/KI-Agenten-diskutieren-auf-Reddit-Klon-Menschen-duerfen-zuschauen-11161385.html | * https://www.heise.de/news/KI-Agenten-diskutieren-auf-Reddit-Klon-Menschen-duerfen-zuschauen-11161385.html | ||
| Line 868: | Line 867: | ||
Die Gefahr im Moment ist eher, dass die unterentwickelten Bots die Bankkonten ihrer Besitzer plündern. | Die Gefahr im Moment ist eher, dass die unterentwickelten Bots die Bankkonten ihrer Besitzer plündern. | ||
Gleichzeitig | Gleichzeitig sind '''reale Gefahren''' eher '''in NAI-Anwendungsgebieten''' zu finden. Beispiele: | ||
* Neueste LLMs und Codex-Varianten können in den Händen von hochprofessionellen Anwendern und staatlichen Akteuren massenhaft [https://www.golem.de/news/claude-opus-4-6-ki-findet-ueber-500-zero-day-luecken-in-open-source-software-2602-205139.html '''Zeroday-Exploits'''] offen legen und beim Hacking und bei der Schadware-Programmierung helfen. | * Neueste LLMs und Codex-Varianten können in den Händen von hochprofessionellen Anwendern und staatlichen Akteuren massenhaft [https://www.golem.de/news/claude-opus-4-6-ki-findet-ueber-500-zero-day-luecken-in-open-source-software-2602-205139.html '''Zeroday-Exploits'''] offen legen und beim Hacking und bei der Schadware-Programmierung helfen. | ||
* Das Muster-Remixing von GenAI gefährdet [https://www.zdfheute.de/wirtschaft/synchronsprecher-boykott-netflix-ki-stimmen-100.html Voice Actor, Synchron-Sprecher] und Übersetzer. | * Das Muster-Remixing von GenAI gefährdet [https://www.zdfheute.de/wirtschaft/synchronsprecher-boykott-netflix-ki-stimmen-100.html Voice Actor, Synchron-Sprecher] und Übersetzer. | ||