# Three Paths to Engage With AI In Knowledge Work By [Dogukan Ozgen](https://paragraph.com/@dozgen) · 2024-02-19 --- We all hear stories about AI and work. and they bounce in a large spectrum – from wiping out all work (yes, including the white-collar ones) and discarding most people out of jobs to creating new jobs for everyone and unseen prosperity. It’s not easy to see what’s to come but one thing is for sure: Like the Internet, AI will change how we work, learn and live like never before. So, how do we navigate this change? That’s the question I chase and I bring some ideas from both economics, tech and education. **2 Levels of Tech Use** ------------------------ [Chris Dixon](https://twitter.com/cdixon) , in his book ‘Read, Write, Own,’ notes that we use new tech in two ways. **Level 1:** In early stages, we do our tasks faster, better and higher quality. **Level 2:** We use it do things that we couldn’t do before. For example, early internet was about reading and accessing more material to read. It was like print-on-steroid. By 2000s, people start to write on the net – ReadWriterWeb (Richard MacManus). That was something new to do… It looks like level 1 use of AI reached mainstream: From summarizing material to translations, analyzing data to internet search – we do our ordinary tasks but in a faster, better way. But, are we tapping into the full potential of this technology? In other words, do we use it at Level 2? I borrowed ideas from multiple disciplines to craft a map for us, knowledge workers. In summary, we can interact with AI with 3 approaches: * Competition * Cooperation * Co-construction Let’s start with the first one, competition, shall we? **Competition: What if AI takes all the jobs?** ----------------------------------------------- You read it in the news, and the reports. This approach positions AI as a competitor. Hence, the goal here is to differentiate – we need to find skills and areas that we can outperform AI. Routine or procedural tasks like coding can be in the wheelhouse of AI. So, our focus should be on tasks that human touch is essential. In a quick sketch, here are some skills and areas to consider: * **Creativity:** Creativity is more than art. As Seth Godin [elegantly put it](https://seths.blog/thepractice/), creativity is about bringing original thoughts. This could happen in tweets, meetings, writings and more. There are many applications of it in our lives. _We negotiate, share and build ideas, and learn from each other._ * **Context-based skills**: Research, program implementations and commerce – all requires an understanding of the context. And, context is not always visible. Also, it changes from where you look at it. One needs to have an understanding of the culture, relations and historical. These factors define the success of a product, a program or a strategy. Otherwise, like Peter Drucker famously said , “culture eats strategy for breakfast.” * **Caring:** Historian Noval Harari [notes](https://www.amazon.com/Lessons-21st-Century-Yuval-Harari/dp/0525512179) that expanded lifespan would cause the rise of the elderly care market. People will want to be taken care of other people, instead of AI-powered robots. You can expand from this point – any professional role in the scope of caring has a leg up agains the machine. While competition have a darker vision blended with automation, techno-optimists offer a brighter version, in which we can do things better, faster, and cheaper. That’s the cooperation. **Cooperation: How can I do this better?** ------------------------------------------ Cooperation is what Dixon described as level 1 – we use tech to do our ordinary tasks in a better way. Today, most of my AI use falls into this category. And there’s a good reason for that. By outsourcing so time-consuming tasks, I open up more free time, increase my output and get faster feedback. My use, as an example of cooperation, are 5 buckets: * **Igniter:** Sometimes you need some help to kickstart your work. GenAI can build an outline, or a broad framework. * **Explorer:** It can help with searching journal articles on a certain topics. For example, Scholar AI saves me hours by helping you find scholarly work. * **Explainer:** When researching a topic, especially in a different or new discipline, the chances are you end up with full of jargon. AI, has innate talent to explain it to you in a simple, jargon-free way. When I was TA’ing for Prof. [Beane](https://www.wildworldofwork.org/)’s “Managing Tech Organizations” class, this use case was the most popular among students. * **Editor:** Many people, including multilingual people like myself, ask for help to check my grammar, editing. Once done, I double check the work and make sure it reflects my voice. This article is an example for it. * **Debugger:** AI is extremely well in reviewing code.  When I end up with an error, I paste the code and ask it to debug. It helps with learning a lot. If you deal procedural (e.g., step by step processes) or technical things, you, too, can benefit from this. As you see in these examples, in the cooperation mode, you allocate tasks in between AI and yourself. This is the basic use case and you can build on it. You can train AI model for your personal needs to get more specific feedback. Currently, I’m training a GPT to help me with my Mixed Methods class. Since it has a background on where my skills at and where I am heading, I get more personalized guidance where to focus on. And,  I am just scratching the surface of possibilities. There are many things we can now do, but couldn’t do before – what I call as level 2 use- and that’s what takes us to the final step: Co-construction. **Co-construction: What can I do with AI that I do that I couldn’t do before?** ------------------------------------------------------------------------------- Constructivism, in learning theory, positions the learner as the builder of learning. Instead of receiving knowledge from outside, we build our knowledge through our interactions with our environment, things and people.  And, our existing ideas are important because we construct our new learnings on them.  With this approach, we can work with AI by building new things upon our combined (AI and ourselves) knowledge and capacities. This way, we can explore into new horizons that we haven’t gone before. **Refik Anadol**’s “[Machine Hallucinations](https://www.moma.org/magazine/articles/821)” is a good example. This dynamic artwork is created by unsupervised machine learning – a form of AI that continuously creating itself -.  It is a blend of artist’s vision and AI’s unique contribution. Before this collaboration, neither AI model nor the artist could create something like this. Now, both of them has some new skills, and knowledge. It’s a good example of Level 2 tech use. / ![Installation view of Refik Anadol: Unsupervised (MoMA)](https://storage.googleapis.com/papyrus_images/f03d9b66972cb4a7593ee6ae98fd510a12ff068e83ba6cb8d898ec37c0c94083.png) Installation view of Refik Anadol: Unsupervised (MoMA) Similarly, we can find ways to use AI models to enhance our work’s limits, explore new destinations. * **Analysis:** AI can analyze the data for a large dataset and humans can provide context and depth with qualitative research and they can create Mixed Method studies. On another field, AI can do diagnosis from medical images and human doctors can provide the case and make customized (nuanced) judgements. * **Creative collaborations:** You can bring in creativity by using AI’s capacity to generate ideas quickly. If you ask for a list of ideas, ask that question again but request a longer, better or weirder list. Once you have set of ideas, you can embed your vision, and add the depth and cultural relevance into it. * **Novel problem-solving/Decision making:** When solving problems, AI can offer multiple perspectives to solve it. Then, humans can decide which ones to pursue and practically applying the solution. Later on, they can feedback the AI and improve the solution capacities. **In Summary** -------------- All these three options (Competition, Cooperation and Co-construction) are not distinctive. You can pick and choose one of them depending on your current task. Or, you can combine them with each other. Perhaps you can start with a competitive approach to clarify what would be your unique input. Then, you would think of which tasks to be delegated to AI (cooperation). After you clarify both your and AI’s parts, it might be good to take a step further and create something new. Here’re some guiding questions to work through these 3 approaches: 1. What you can thrive on? 2. What you can do better (e.g. , faster, cheaper, easier, or higher quality)? 3. What else can you do (e.g., something you couldn’t do before)? _And some more…_ As I get to finish this article, ChatGPT suggested add a few more thoughts and I turned them into questions as well: * What are the ethical implications of my use? * Can there be any privacy and security breaches (e.g., mine or others) in this use? * Are there implicit biases in this use? If so, how can I make them explicit, or counter them? I hope this helps your adventures with AI. --- *Originally published on [Dogukan Ozgen](https://paragraph.com/@dozgen/three-paths-to-engage-with-ai-in-knowledge-work)*