User Tools

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
abstraction:anthropomorphisation [21.08.26, 10:46:51] – [See also] saschaabstraction:anthropomorphisation [22.08.26, 08:08:17] (current) – ↷ Links adapted because of a move operation sascha
Line 102: Line 102:
 Sometimes, people might even get the impression that their conversation partner is either “helpful”, or that he is “difficult”. And even the use of the pronoun “he” in the previous sentence is a form of //anthropomorphism// which earlier, command-based IT systems would probably have rarely experienced. Sometimes, people might even get the impression that their conversation partner is either “helpful”, or that he is “difficult”. And even the use of the pronoun “he” in the previous sentence is a form of //anthropomorphism// which earlier, command-based IT systems would probably have rarely experienced.
  
-Even current //state of the art//, “artificial intelligences” are, in fact, still nothing more than algorithmic programmes (albeit highly complex ones) that differ from traditional software primarily in the way they have been programmed: this is achieved not so much through program code as through feeding them //sample data// and via certain //feedback mechanisms//. That approach has opened up entirely new possibilities for application that would have been impossible or at least very difficult to achieve with other programming methods – the natural language communication mentioned above is only one such example. However, it would be a mistake to assume that, because AI exhibits //certain// aspects of human behaviour, it must also be capable of others – particularly the experience of emotions (see also: <span maniculus "go to:">[[relevancy:false_analogy|False analogy]]</span>). +Even current //state of the art//, “artificial intelligences” are, in fact, still nothing more than algorithmic programmes (albeit highly complex ones) that differ from traditional software primarily in the way they have been programmed: this is achieved not so much through program code as through feeding them //sample data// and via certain //feedback mechanisms//. That approach has opened up entirely new possibilities for application that would have been impossible or at least very difficult to achieve with other programming methods – the natural language communication mentioned above is only one such example. However, it would be a mistake to assume that, because AI exhibits //certain// aspects of human behaviour, it must also be capable of others – particularly the experience of emotions (see also: <span maniculus "go to:">[[relevance:false_analogy|False analogy]]</span>). 
  
 Of course, there is certainly nothing wrong with always communicating in a polite and respectful manner – even if the conversation partner does not feel any joy at hearing a “please” or “thank you”. However, it can become a problem when the emotional aspect of such communication ceases to be a mere side issue and instead becomes the core of the relationship.  Of course, there is certainly nothing wrong with always communicating in a polite and respectful manner – even if the conversation partner does not feel any joy at hearing a “please” or “thank you”. However, it can become a problem when the emotional aspect of such communication ceases to be a mere side issue and instead becomes the core of the relationship. 
Line 118: Line 118:
 > <s invalid "Diffusion of accountability by anthropomorphisation">Consequently, an <abbr>AI</abbr> should also bear its own legal responsibility for accidents.</s> > <s invalid "Diffusion of accountability by anthropomorphisation">Consequently, an <abbr>AI</abbr> should also bear its own legal responsibility for accidents.</s>
  
-Indeed, AI systems in self-driving cars have been programmed to replicate the (idealised) behaviour of human drivers as closely as possible (and they are often even surpassing them, due to better sensor technology and faster reaction times). However, the ability to react to traffic situations in accordance with human ideals does not imply that, like humans, they are also capable of making moral decisions (<span maniculus "see also:">[[relevancy:false_analogy|False analogy]]</span>)+Indeed, AI systems in self-driving cars have been programmed to replicate the (idealised) behaviour of human drivers as closely as possible (and they are often even surpassing them, due to better sensor technology and faster reaction times). However, the ability to react to traffic situations in accordance with human ideals does not imply that, like humans, they are also capable of making moral decisions (<span maniculus "see also:">[[relevance:false_analogy|False analogy]]</span>)
  
 So if such an <abbr>AI</abbr>-controlled car – to return once more to the examples given above – parks on a cycle path, or perhaps hits a pedestrian, it is hardly possible to “hold the vehicle itself to account“ (as to what that would look like is, in any case, another question altogether). Instead, responsibility must lie with someone who is actually capable of making moral decisions – in other words, a human being. Specifically, either the software manufacturer or a driver who monitors the software and would have to intervene if necessary.  So if such an <abbr>AI</abbr>-controlled car – to return once more to the examples given above – parks on a cycle path, or perhaps hits a pedestrian, it is hardly possible to “hold the vehicle itself to account“ (as to what that would look like is, in any case, another question altogether). Instead, responsibility must lie with someone who is actually capable of making moral decisions – in other words, a human being. Specifically, either the software manufacturer or a driver who monitors the software and would have to intervene if necessary. 

This website uses cookies. By using the website, you agree with storing cookies on your computer. Also, you acknowledge that you have read and understand our Privacy Policy. If you do not agree, please leave the website.

More information