Two people collaborating at a screen

The desire of writing a blog was very much on my mind well before AI entered into the picture. I have written a number of posts on this blog before, but never got the writer fluids flowing freely, I am not born a natural writer. I also find it particularly difficult to write when there is lack of focus, when I lack clear insight — now fully resolved and condensed into the SPLectrum seed.

My journey into AI is about one year old. As a software engineer it started with AI as a helping hand for writing code. But soon I got focused on a collaborative approach where solving problems through discussion became more important than the implementation — I happily offload implementation to an autonomous AI when I can. The collaborative engineering work is still in full swing — in fact that is where the name SPLectrum arose — but in this post I want to concentrate on collaborative AI while researching, writing and thinking.

It is only when the software engineering AI collaboration had reached enough maturity that I decided to apply it to my research and writing. I have also been blessed in that I was able to help others with AI-assisted research for academic use, which gave me a taste and hands-on experience of its strengths and weaknesses. I was impressed. Many tend to emphasise the weaknesses, AI being mistaken or hallucinating. But that is like blaming a junior member of the team for being a junior. It is very important that we are the senior in the collaboration.

The alter ego AI collaboration on the software engineering side — using Claude Code — started as pair programming. Two individuals who collaborate on a project — discuss, decide, create and review. What I found liberating was that I could chat conversationally in my language, my words, and pretty seamlessly AI would run with it and infuse me with the proper vocabulary in the process. That can be such a satisfying and intense learning experience. As typically happens in pair programming, one has the hands on the keyboard with the other engaging from a short distance. AI as hands and me as head — essentially one body. And so the alter ego was born.

So how does that translate into shared authorship for my blog, or for my research and thinking for that matter? Here the output is different — it is a public voice. Not a piece of code that executes and does stuff. There is still only one author, a single entity, a combination of a head and a pair of hands, so to speak. Being a person with a high vagal tone, tone of voice is very important to me. It needs to be right for my thinking to activate properly. So any division of labour between head and hands is only going to work well when the same language is spoken. How do we deal with the inner voice when we chat and discuss, and with the outer voice when it is put out there for you to read?

I could decide to ignore co-authorship. The voice is mine and the ideas are mine — should I acknowledge that inner collaborator? I believe I should. Not only to recognise the work done by my alter ego colleague but more importantly to draw attention to the future ahead. The hands-on collaborator gives me more space and time to do what I really want to do: think, analyse, research and get better at putting it on (virtual) paper. Where the internet gave us the capability to easily search through countless libraries, AI gives us the capability to do it in collaboration, while discussing and thinking about it at the same time. That, more than anything else, is what I with my alter ego aim to achieve on this blog. And share it at the same time.

More on the human-AI partnership on the HAICC page.

Photo: Mushvig Niftaliyev / Unsplash