Please login or join to subscribe to this thread
It has no sense at all to debate about it. Those type of things are in the imagination of people. The final decision when an AI device (software or hardware) has analyzed somthing and presented the results of the analysis is still on hands on human beings. So, go to the human being. Does somebody decide that the final decision skip the human being and will be shut by the AI device? Yes. But is the same when somebody decide the existence of the red buttom to fire the nuclear bombs. In my country we said "son locos, pero no comen vidrio" trying to translate it into english "they are crazy but they don't eat glass". I am working and researching in AI from 1989 in several fiiels (health care to military). I know about I am talking about but thanks God I do not own the truth.
Ethics is a human construct. AI has no concept of ethical vs. unethical behavior. Depending on how that AI has been constructed, it will depend on how ethically it will operate. For example, if machine learning is used to construct a job screening algorithm and the data set used is biased towards or against certain groups, then the outcome will be unethical.
Once AI becomes more capable and edges towards sentience, we may need something similar to Asimov's three laws of robotics...
In my view, the algorithm is an unlikely source for bias in modern weak-AI approaches. These applications are metadata-driven, and the cognitive biases are implemented through the humans that manage and/or own the metadata.
To state it differently, the “world view” of the metadata managers is what makes me shutter. These managers will state they have configured the system per studied and agreed upon facts, but those facts (i.e., raw data) are still interpreted under world-views and then cast into metadata that then drive behavior.
If you want true super powers, then find a position where you are the one casting knowledge into metadata that controls the behavior of a massively consumed service. That is the unfortunate truth and that which defines the current and future clandestine power brokers.
Thank you, gentlemen, for your perspectives and insights into this murky world of AI, ethics and biases.
Thanks for sharing this article. Ethics & AI will become a hot topics
Elon Musk mentioned the need for an oversight body that would regulate Machine learning and AI practices. The need for regulating new technology is very prominent in 3D printing for example as Ethics are very subjective
Agree with Fabio, ethics & AI will be an emerging topic once it becomes more ubiquitous.
I would first look at the legalities of what can be supported by AI and how the legal system may interpret AI.
The legal system does not need a law explicitly stating the uses of AI and what it can and cannot be used but instead makes ruling based on existing laws that protect an individuals rights.
If it sounds ethical dubious then there is probably a law that can be inferred that does not allow this practice to occur.
I have a concern with the your sentence ""If it sounds ethical dubious then there is probably a law that can be inferred that does not allow this practice to occur"
Although an existing or new law may stipulate that certain practices should not occur and that law can be and will be enforced it will not prevent the practice from occurring. All a law can do is make it illegal.
I have a problem with your sentence " All a law can do is make it illegal...it will not prevent the practice from occurring".
If as a project manager you do not abide by the laws of a country then anything you touch will be compromised and anyone you associate will be have their reputation tarnished as a result.
The law does not legislate for common sense, doing good or being a good citizen but builds and designs a framework that is flexible, practical, easy to understand and easy to apply.
For example in AI if you are replacing a person with a chat bot you must inform the person at the start of the conversation that you are indeed chatting with a chatbot. This is a legal requirement not a ethical decision yet their is no law in many jurisdictions that require this as it has been inferred from other laws on the statute books.
Please login or join to reply