Most people had not paid too much attention to AI until recently. While the simulation of human intelligence in automating boring household work is still low in our lives (to my greatest regret), AI appeared in different professions overnight and started to replace knowledge-intensive activities such as targeting online advertisements, or personal assistance. It turned out, that on one hand, manufacturing physical robots that use tools designed for humans seems to be more challenging than any science fiction author would have expected, on the other hand, some of the intellectual work can be surprisingly easily mimicked by algorithms and consequently learned by AI.
One of the most common application areas of AI is natural language processing, which focuses on teaching machines to understand and interpret human languages. Not so long ago, we had to look up unknown words in a printed dictionary, then online dictionaries accelerated the process significantly, but now a translation app, like deepL, does the whole job for us in a glimpse. The real superstar in this field, that turned our attention to the possibilities of AI, is the free-of-charge ChatGPT. For those, who haven’t heard about Chat GPT: It is an online chatbot that has been trained on a large corpus of text data, which enables it to generate high-quality responses to a wide range of topics. It is open-source and keeps improving. Probably scientists working in this field will not be impressed too much by this service, but for the general public, it is a kind of magic; an easily accessible app that can talk to us and solve versatile problems. No wonder it has received a lot of attention recently. Even the mainstream media is full of its innovative or surprising application areas from writing books, poetry, or student essays to authoring press releases, and generating movie plot lines or game dialogs with narrative content.
But can it help organizational development? Does it have relevance for us, consultants, and for the management? After a bit long introduction, this blog post aims to answer these questions through some examples and my initial experiences with this remarkable application.
Fields of application in organizational development
First of all, ChatGPT can be a great tool to support organizational learning. As an HR manager or learning and development consultant, you can ask it to map out learning goals on any given topic and create training materials. It can provide insights and knowledge related to various topics, such as leadership, teamwork, communication, organizational culture, and change management. It can write an eLearning script, a guideline for a panel discussion, or a branching scenario too, just to mention some concrete examples.
Second, there is a vast potential in internal communication for ChatGPT. It can write announcements and speeches to increase engagement, clarity, and transparency within the organization. Furthermore, it can help in developing textual elements of an effective communication strategy, or even delineating the entire strategy. Although right now the natural language responses might not be completely satisfactory in case of such word-selection sensitive or complex user questions, we can assume, that ChatGPT will be able to generate soon the required content fast and accurately.
Third, ChatGPT can assist in conducting surveys and data analysis. It can write survey question alternatives for a skills-gap analysis or propose questionnaire items to measure employer satisfaction. In addition, it can help in encouraging survey participation, making suggestions related to data collection and analysis, and can propose ways to present the data or summarize the results.
Personal impressions:
I tried ChatGPT for various purposes during the last few weeks (from writing internal communicational materials to mapping learning goals in a given topic), and here are my initial general impressions:
- In most of the cases, I received relevant and more or less correct responses to my questions, but not always. ChatGPT is not infallible, so a second review or double-checking of the facts and the literature can be necessary before we take the responses too seriously.
- I missed the references to the literature/documents/websites it had used to generate the answers. I deem it important, where the information comes from or what legislation is the basis of its recommendation. When you ask ChatGPT to recommend literature on a given topic, it lists papers or books, thus, it solves the problem, however, relying exclusively on the answer can cause a one-sided or misleading approach.
- In most of the cases I had the feeling that during generating the answer, the aim was to create a lot of text rather than answering the questions in substance. I asked the very same question from a professional and from ChatGPT (I tried it with more questions and more subjects), and the main difference was every time, that the professional was more to the point, more practical, and more precise in reasoning or referencing.
- It is almost unbelievable that the software has no problem talking to us in other languages than English. I’m Hungarian, and my general experience is, that automated language translators or speech recognition do not work very well in my mother tongue. Although also ChatGPT does grammar mistakes, its performance is very high.
- Strategic and critical thinking has major importance when it comes to ChatGPT. You have to know what are the good questions or commands and be able to process the answers you receive in an appropriate way.
- ChatGPT can help you in starting your work or give you extra discussion points, but it won’t do your job. At least for now. I asked ChatGPT to write this post for me, then I flirted with the idea to use it as a co-author, but finally, I decided to write my own thoughts, and did not use a single sentence from the answers it gave me to my questions. So I wanted to speed up the work at least, but in fact, even his background research was incomplete. So much about replacement and copy-and-pasting.
In sum, we can say, that in spite of its deficiencies and (probably only initial) limitations, Chat GPT has already a huge potential in good hands. However, on the other hand, it also raises many questions related to ethics, disinformation, bias, or social consequences. Talking about the complete replacement of repetitive human jobs, ChatGPT can be a game changer, despite still having teething problems. Remember that one of the characteristics of artificial intelligence is its ability to learn, so it is constantly improving itself. ChatGPT or similar software will soon be able to provide relevant communication materials or give advice on well-formulated OD questions.
Will knowledge from texts available on the Internet be a substitute for real-life experience? That is a good question for the time being. ChatGPT received many criticisms from educators, journalists, artists, and public advocates, calling it a “stochastic parrot” (Hengel), a tool for disinformation and manipulation, or a source for many already visible and unforeseeable dangers.
Although we might have the feeling, that we have opened Pandora’s Box with ChatGPT, it is not even the only AI solution in our hands. Here are some examples that I haven’t tried yet, but saved for later: Excelformulabot.com transforms text instructions into Excel formulas, SlidesAI.io generates presentation slides in multiple languages, midjourney.com creates images from text, or durable.co build websites based on user-given information. We can expect, that as the field continues to grow, new subfields may emerge and some subfields may become more and more specialized. ChatGPT is only the start, and it’s up to us what we do with it. The only one what we cannot do right now is to leave these potentials out of consideration.