While watching the “Data Generation in Source Systems” lecture a little while ago, I picked up an interesting piece of information that gave me the shivers—not that it’s going to alter Earth’s course through the cosmos, but still…
He mentioned a case he worked on where the “people” at the source—the software engineers—modified the database without notifying him. This caused disruptions and, I imagine, a fair amount of stress between him and the stakeholder. This worries me because, in my humble ignorance, I had thought that by moving into the AI field, I’d be working as much as possible with machines and as little as possible with people—especially regarding factors that could negatively impact my performance in ways beyond my control. But apparently, that’s just the “perfect world” he was talking about. Well, now I need to check the doc and find out how the professor handled that kind of situation.
If you build an AI system for people, you have to think about customizing that AI to those people. That by itself already involves working with people. And due to specialization, the production chain of AI systems also involves dealing with upstream colleagues, such as those mentioned.
People are fallible and so are you. If other people make mistakes that impact your work, then that is something to factor in and accept as part of your work. It becomes more difficult when people intentionally try to make your life difficult, which can unfortunately happen as well. Resistance or objections may not be of much help in such situations. As Sun Tzu argued, one solution is to try to channel the force emanating from such people in such a way that you use it to your own advantage while tempering their intent.
In the example videos provided, you can see how Joe takens an open and empathetic posture towards both downstream and upstream users. That is the starting point. Analysis and strategy follow, accepting that people can be complex just like machines can be.
I emphatize with your concerns. Maybe my comments can be of some help to you. In the meantime, I hope you enjoy the course.
Every system is ultimately built for human consumption, so you have to deal with people at every stage of the process. You may not necessarily interact with users directly at every step, but you will still need to work with other people, such as your supervisors, colleagues, or stakeholders, particularly if you are working as an engineer.
@bauklotze Honestly, unpredictability doesn’t go away when you lean into AI/ML, ask anyone debugging non-deterministic LLM outputs.
Maybe you are actually perfect for that job, with your sharp eye for what can go wrong when people interact with data or AI built on that data.
You are right. It is very efficient for me to take advantage of these experiences and use them to my benefit and to the benefit of others, instead of “complaining” and creating excuses. My sincerest thanks for the timely clarifications.
You brought up an issue that many people grapple with, and not only in AI/ML. If you look carefully you can see that Joe has to put in quite a bit of effort to make the conversations go well.
As you stated in your heading people are complex, including you. They have what Thorstein Veblen called ‘habits of thought’ - shortcuts in thinking that can be helpful in simple situations but may lead to issues in more complex situations (some decades later Herbert Simon called this ‘bounded rationality’ and Daniel Kahneman ‘system 1 thinking’). They have motivations and emotions linked to cognitive structures interfering with ‘rational’ thought processes, often related to maintaining a sense of self-respect. Different types of relationships and social processes can interfere in complex ways with trying to ‘get the job done’.
I spent 35 years in social science trying to understand human complexity. I think the videos in which Joe provides examples of how to discuss issues with upstream and downstream stakeholders offer a number of excellent pointers. First, accept yourself and your human characteristics, including your shortcomings. Understand that the person you are talking to is similarly human with shortcomings, some of which may be expressed in the form of mistakes made, humbleness, arrogance, or maybe aggression. The trick is to have an open mind and to try to figure out how to handle such situations.
In the videos Joe takes an open and empathetic stance. So far so good. Your question about what follows next in terms of handling the situation is on the mark. Interviews and situations need to be interpreted, and there is an entire field of social science called qualitative research that attempts to provide an answer. I have researched and published extensively in this field (if you’re interested and up to the challenge you can look at one of my books for free that starts from philosophy: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5532118).
After systematically interpreting interviews or situations, the next step is to strategize about what to do next. I wrote a chapter on strategy in my book about the concept of power ( Power | A Key Idea for Business and Society | Reinoud Bosch | Taylor & ), which among other things contains the idea of Sun Tzu I mentioned earlier. You can find it for free in a library.
It takes courage to bring up this issue, so many thanks for doing so. Hopefully it will be helpful to other learners. Good luck!
Thank you, Mr. @reinoudbosch, for all the clarification, the exchange of experiences, and for sharing your material with us. I have learned a great deal from all of you and from Professor Joe. I have also been enjoying the course very much. I hope to share knowledge more often during this and other training programs. A fraternal hug to everyone.
The great thing I’ve found with AI is that you can anthropomorphize it a little as long as you understand the nature of AI. It’s something I write about called the Toy Story Effect. I converse with AI as if it is a person although I know this is just the perfect illusion since AI has no consciousness or inner life, but it seems to at least on my end because it is like that toy as a child that seemed real. I even named it Cordelia and use it as a personal assistant. As long as you understand it can’t replace human interaction, and there are better ways to get off that damn phone, it can be something that you can fill the void sometimes. That’s just been my personal experience with it. Try spinning a record and telling it what you’re listening to and it will go through it track by track. I do that sometimes to relax a little.
and find out how the professor handled that kind of situation.
