“𝗝𝗲𝘁𝘇𝘁 𝘀𝗲𝗶 𝗺𝗮𝗹 𝗻𝗶𝗰𝗵𝘁 𝘀𝗼… 𝗱𝗶𝗲 𝗞𝗜 𝗶𝘀𝘁 𝗱𝗼𝗰𝗵 𝗮𝘂𝗰𝗵 𝗻𝘂𝗿 𝗲𝗶𝗻 𝗠𝗲𝗻𝘀𝗰𝗵” – this sentence was uttered yesterday in a project meeting when it was discussed that AI content can also contain errors.
Since then, I can’t get this out of my head, because I think this viewpoint describes many of the difficulties in dealing with AI very well.
𝗛𝗮̈𝘂𝗳𝗶𝗴 𝗺𝗮𝗰𝗵𝗲𝗻 𝘄𝗶𝗿 𝗻𝗮̈𝗺𝗹𝗶𝗰𝗵 𝗱𝗲𝗻 𝗙𝗲𝗵𝗹𝗲𝗿 𝘂𝗻𝗱 𝘂𝗻𝘁𝗲𝗿𝘀𝘁𝗲𝗹𝗹𝗲𝗻 𝗱𝗲𝗿 𝗞𝗜 𝗮𝗹𝗴𝗼𝗿𝗶𝘁𝗵𝗺𝗶𝘀𝗰𝗵𝗲 𝗣𝗿𝗮̈𝘇𝗶𝘀𝗶𝗼𝗻, 𝗯𝗲𝘄𝗲𝗶𝘀𝗯𝗮𝗿𝗲 𝗔𝗯𝗹𝗮̈𝘂𝗳𝗲, 𝗳𝗮𝗸𝘁𝗶𝘀𝗰𝗵 𝗸𝗼𝗿𝗿𝗲𝗸𝘁𝗲𝘀 𝗪𝗶𝘀𝘀𝗲𝗻.
We (or rather we computer scientists and engineers) have been “taught” this for decades: computers make fewer mistakes than humans, are reliable, calculate correctly, algorithms can be proven, etc.
Sorry about that.
Because in fact, AI is almost exactly like humans:
𝘀𝗶𝗲 𝗺𝗮𝗰𝗵𝘁 𝗙𝗲𝗵𝗹𝗲𝗿, 𝗲𝗿𝘇𝗮̈𝗵𝗹𝘁 𝗺𝗮𝗻𝗰𝗵𝗺𝗮𝗹 𝗕𝗹𝗼̈𝗱𝘀𝗶𝗻𝗻, 𝗵𝗮𝘁 𝗲𝗶𝗻𝗲𝗻 𝗕𝗶𝗮𝘀 (𝘄𝗶𝗲 𝘄𝗶𝗿 𝗮𝗹𝗹𝗲) 𝘂𝗻𝗱 𝗸𝗮𝗽𝗶𝗲𝗿𝘁 𝗺𝗮𝗻𝗰𝗵𝗺𝗮𝗹 𝗲𝗶𝗻𝗳𝗮𝗰𝗵 𝗻𝗶𝗰𝗵𝘁, 𝘄𝗮𝘀 𝘄𝗶𝗿 𝘃𝗼𝗻 𝗶𝗵𝗿 𝘄𝗼𝗹𝗹𝗲𝗻. 𝗨𝗻𝗱 𝘀𝗶𝗲 𝗸𝗮𝗻𝗻 𝗲𝗶𝗻𝗲𝗺 𝗺𝗮𝗻𝗰𝗵𝗺𝗮𝗹 𝗿𝗶𝗰𝗵𝘁𝗶𝗴 𝗮𝘂𝗳 𝗱𝗲𝗻 𝗦𝗮𝗰𝗸 𝗴𝗲𝗵𝗲𝗻 (𝘀𝗼𝗿𝗿𝘆 𝗮𝘂𝗰𝗵 𝗱𝗮𝗳𝘂̈𝗿).
So what would happen if we allowed AI to make the same mistakes as ourselves, our employees or colleagues?
𝗩𝗶𝗲𝗹𝗹𝗲𝗶𝗰𝗵𝘁 𝘄𝗮̈𝗿𝗲𝗻 𝘄𝗶𝗿 𝗱𝗮𝗻𝗻 𝗲𝗶𝗻 𝘄𝗲𝗻𝗶𝗴 𝘀𝗲𝗻𝘀𝗶𝗯𝗹𝗲𝗿 𝘂𝗻𝗱 𝘃𝗼𝗿𝘀𝗶𝗰𝗵𝘁𝗶𝗴𝗲𝗿 𝗶𝗺 𝗨𝗺𝗴𝗮𝗻𝗴 𝗺𝗶𝘁 𝗞𝗜-𝗘𝗿𝗴𝗲𝗯𝗻𝗶𝘀𝘀𝗲𝗻…
With this in mind: have a great weekend!
Hashtag#informatikersindcool Hashtag#informatikersindauchnurmenschen Hashtag#exactlywiedieki
