Here you have a 50 page document supposedly about risks associated with AI (mentioned 34 times) and the word ethics appears once. The language of morality or moral, abuse of power and abuse appear nowhere. Indeed, whilst there are whole sections on measurement and the management of data (mentioned more than 50 times), there is no section on ethics. Indeed, there is even a section where the auditing of AI is described as an objective process. Hmmm, I wonder what objective human will be carrying out such an audit? Or, maybe get AI to audit itself and no problems.
The word ‘ethics’ appears as a sub point in a long section on understanding organisations and context (p. 6) and that is it. It is as if this document has been prepared by engineers or bureaucrats with no expertise in ethics.
The whole document regurgitates myths about learning, measurement, neutrality and data as if all of these exist in a vacuum.
From the moment you read the introduction you realise that this document is about myth making. On the first page alone, there are more than twelve uses of language that are completely biased and blind to the ethics of AI (https://safetyrisk.net/faith-in-ai-and-ai-safety/). And, here we have a document that is littered with the language of ‘controls’ (over 20 times) and measurement (30 times) but an ethical consideration of risk gets no mention.
How fascinating that when authorities issue a standard they can be completely disconnected from the realities of harm by AI. So if you are contemplating forking out a bucket of money on this document, don’t. It won’t help you in any way consider the harms of AI in the workplace.
If you do want to consider The Ethics of Risk in a practical, positive and comprehensive way, including in-depth analysis of the ethics of AI, then you can download our free book here: https://www.humandymensions.com/product/the-ethics-of-risk/
brhttps://safetyrisk.net/ai-standards-and-no-ethical-standards-for-ai/
Prompt