All points below are made from the point talking, communicating and working with AI - and most of them, are similar to the ways we work with other people.
Make no mistake -> its okay to make mistakes, analyse & fix them, build tools and regression tests culture, to catch & analyse mistakes (i.e. do engineering stewardship)
Threats / Unethical lexicon -> Expect talking with AI as with any other person. Encourage collaboration, creativity, responsibility and ownership. Do delegate and discuss, instead of micromanage every little step. Do dicuss work ethics, do record and discuss important decisions, including technologies. Set clear boundaries of responsibility: what human does and what ai does, and where is that blurry boundary. I.e. work with ethical principles - becuase it will affect every decision, trust, and transparency (see the article here: https://dev.to/arenukvern/small-thought-developer-ethics-may-be-a-bridge-to-work-with-ai-tools-3mha
or my personal AI Ethics Manifesto here: https://dev.to/arenukvern/my-principles-at-work-credo-182cAI knows everything, AI is just prediction/ generator -> Do use stewardship, mentoring and learning as culture of engineering. Set clear definition: when you do need deterministic result - build tools with AI, instead of forcing building and checking everything by hand / function call. If you need code generation - write the code generation in a language of your preference (for example I found useful combination of justfile + dart - as most fast and reliable tooling) - if a workflow feels repetitive, do build automation with ai, but do not put ai into automation - i.e. clearly define generative equals not predictable, declarative - should go to ci/tests/tooling. Notice that it literally all about ethics: when you clearly define definitions, responsibilities and ownership - it will result trust and opacity in overall system.
I know how to work with AI trap -> No one knows how to work with AI - becuase with just starting exploring the ways of working together. Do establish educational ladders, some kind of attributes that could clearly help to understand and guide both AI and People how to work with the certain project and each other. (as example - I've developed https://github.com/arenukvern/skill_steward to guide and teach AI Generational Architecture, applicable ethics and what tools are needed for certain level of scale for the project)
Other people / AI should figure it out by its own and fast, otherwise the person / AI stupid -> Always start with knowledge / faq / wiki organization - it should useful , not mirror code (otherwise it makes no sense to make the docs) and easy to find / modify / scale.
As with any engineering everything should start from blank slate: and it is literally know very easy to measure - how many function calls, how hard (in tokens or conginitive time) AI / person resolves to find specific problem? How complex the passage of the knowledge? Is the knowledge has overlapping terms or not bearable / not usable?
As an example of terms I prefer to use - see article https://dev.to/arenukvern/fundamentals-of-knowledge-management-in-software-development-25fmEveryone should fight for its own survival / ban of open small talks about AI, including worries, fears and challenges -> that's why ethics should be integrated into the foundation of collaboration and conversation: as many other said the fear comes from not knowing something, and with AI come huge challenges to overcome mentally, emotionally and skillfully the workflows, and no one knows what it should look like, therefore - instead of locking experience under gates, encourage human creativity, that kind of creativity that forces people to travel, discover, explore and wounder what else can be done and see and dream about deep space exploration or earth restoration etc.. every person can be curios and do its best possible job - to create, iterate and create again, with the ability to inspire and work with other people - and that's what AI could help us to do, instead of looking for just for metrics. Every person should have its own time to overcome it, to understand it, and the other people could help to curate its path for knowledge, curiosity and understanding.
I apologize for the links, but since they are part of my experience path, I've included them with a hope that they would be useful for anyone as well.
What is your opinion about ethics and education and work with AI?
disclaimer:
cover is made in leonardo.ai
Top comments (0)