From the perspective of one who has lived through 50-years of computing and software evolution, the dangers of generative AI will not come from the the technology itself but the frailties of those humans, who either fear or misuse it.

It is a shame that the terms "artificial" and "intelligence" were ever coined for it. It is neither.

In the early 1990s I was leading R & D at a large independent software vendor. I invested substantial resources in researching and experimenting with AI as it then existed. At the time, neural networking showed great promise, but was hampered by the limited processing power available.

Times have changed and both the capabilities of neural networking and training algorithms, and the availability of economical processing power have increased exponentially. Coupled with advances in natural language processing, generative AI is now a useful tool.

But generative AI is not magic and it's only a tool. It is neither intelligent nor creative. As a software developer I primarily use ChatGPT-4 as an exceptional search tool. Kind of a Google Search on steroids. Just like any conventional search engine, there is no guarantee that responses are either vetted or accurate. That is the responsibility of the, hopefully intelligent, human using it.

I am also learning to use ChatGPT-4 to generate code that invokes APIs of open source Java libraries (probably the only kind of libraries to which it has access). From what I've seen so far, it would be more work to describe a complex application algorithim to it than to write it oneself. It is certain that app development tools will soon exploit generative AI to make its capabilities commonplace.

Coding is the easier part of developing software. Design is much more difficult. I think that it may take a while before generative AI will be able to understand the functionality, semantics, constraints, and invariants of an application sufficiently well to decompose it into an effective component model with appropriate communications. That will happen eventually, but like a mediocre designer, not necessarily a good one.

The real danger is in the misuse of generative AI through ignorance and/or greed.

For the forseeable future, generative AI is utterly incapable of any kind of creativity, judgement, or moral sense. Honest, prudent human beings will treat it accordingly. My fear is misuse of generative AI by the those other ignorant, greedy, or dishonest human beings.