So I've been full-time independent for 12 years now. In general, I don't think AI is a major driver of employee vs contractor decision but we'll see.

The market has been trending towards specialists for a long time. AI may help employees in the short term be more effective generalists, and so be able to compete with specialists. AI may help specialists be even more effective in their niche, while also serve wider needs, and so compete better with employees.

Something I do see happening is companies are doing a lot of low hanging fruit themselves in my space (I do revenue and business analytics). Today, they will get 80% of my specialty done themselves. That is enough for most companies. But that last 20% for those who want it, still requires a specialist like me who knows the domain entirely.

It’s going to be a great time to be an entrepreneur, but a terrible time to be an employee.

A solo indie developer can have an AI team working on a project.

But so will every corporation. The competition for the remaining human-only roles will be intense.

I think that AI will let us make better progress outside of our narrow expertise.

So entrepreneurial activities will be easier and more common. On the other hand, there will be relatively fewer opportunities for specialized consultants. Contractors and consultants should be able to solve bigger problems rather than working in narrow specialties.

Teams and companies should have fewer members, since fewer specialties are needed. So they will probably need more contractors to move things along when there is a lot of work to do.

The current job market is definitely increasing the amount of freelancers and indie developers.

No I think it will create more full time roles, either (pessimistic) cleaning up AI slop, or (optimistic) opening up work that would've been uneconomical before.

The software crisis of the 1960s was marked by an inflection point: Organizations saw the benefit of automation but software writing was a tedious affair typically done in assembly language or FORTRAN or something. There just weren't enough programmers on earth to take up the load of writing all the software that would be necessary. So new tools, like COBOL and ALGOL, were devised to help programmers produce correct software quickly.

Today's software crisis is not one of too little but too much. We are absolutely spoiled for computing power -- a smartphone having enough capacity to replace a mainframe that in the 1970s or 1980s would have handled a national bank's transactions, many times over. We are awash in software, most of it bad. We need less software and better software. Stochastic slop generators are going to make this problem worse, not better.

It may be a rough few years, but on the other side there will be a boom in demand for programmers to clean up the mess "AI" has made.