Coping with Technology
How do we cope with rapidly changing technology, particularly AI?
For years we have made costly mistakes in complex projects, where we have underestimated the cost, as well as the time to implement what we have proposed. This is a very common problem. The drivers being our hurry to do something, as well as our ignorance of what it takes to achieve what we want as the outcome. There are many examples, the Snowy Mountain Project is a current example, with the current cost exploding to 10 times the original budget, as well as a delivery timeline that cannot be confirmed. There are plenty of other examples in commercial and Public projects.
The processes used to achieve a successful outcome are well known, but not regularly utilised. Why? Well, the most common source is the inexperienced management, not prepared to take the Project / Engineering advice they are given, when it becomes apparent the cost and schedule are not what they expected, so they plough on despite the red flags. The pragmatic manager would take the Project / Engineering advice to rescope the project to meet expectations, however, this is not common, especially when politics is a driver. In failed projects it is always surprising on how many understood the real problem, but have been unable to influence the decision makers, so human nature is the culprit.
Does technology such as AI help? Yes and no. It will if the AI messenger is seen as someone management is prepared to trust. However, it will not if the AI tells you what you want to hear, or conversely helps you to be more efficient at failing.
So, what do we do with our AI? Well, we treat it as a useful assistant not the lead. My own experience with AI is that it can make me very productive, as long as, I know what I am doing.
Systems engineering was created in the 60’s to cope with large project complexity; the US Space program was the testing board and was quite successful. At its centre is the Problem Statement, or the Requirement. It must be clear and unambiguous. The AI equivalent is the Prompt, you can also set up rules to act as guard rails. So AI can be very effective if managed; in my own case I will create a solution functional architecture and work with AI to break this down into modules. AI is great at, User interfaces, API’s , databases, and documentation, and writing any code that is well specified; it is also excellent at reviewing code written by others.
So dwell for a moment; The human, me, is the architect / designer; the AI is my capable assistant, multi-skilled and a productivity accelerator. The decisions are made by me; I also find many errors that go unseen by AI; this can be frustrating, but confirms the need for outcomes to be checked. To be fair the AI also catches me out quite regularly, often in the enforcement of a design rule I had insisted on. So a team effort is involved.
The benefits of AI in capable and experienced hands are immeasurable, the productivity improvements quite astounding in the 5 to 10 times range. The quality of the outcomes are also very high.
