[This blog post has been co-authored with ChatGPT and DALL•E 2. Please see below for a discussion of how this process went.]
In today's fast-paced business world, many companies are looking for ways to stay competitive and achieve their goals. One approach that has gained popularity in recent years is management by objectives, specifically Objectives and Key Results, or OKRs.
OKRs are a way for organizations to set and track their goals and objectives. The idea is that by clearly defining what needs to be accomplished, and by whom, everyone in the organization can work towards a common goal. This can help to improve efficiency and productivity, as well as to align the efforts of different departments and teams.
At its core, an OKR is made up of two parts: an objective and a set of key results. The objective is a clear and measurable goal that the organization wants to achieve. The key results are the metrics that will be used to track progress towards that goal. For example, an objective might be to increase sales by 10% over the next quarter, with key results being the number of new customers acquired and the total revenue generated. While OKRs can be a powerful tool for achieving organizational goals, there are also some potential problems to be aware of. One issue is that OKRs are often mandated from above, and may not reflect the day-to-day goals and priorities of the team. This can make them feel like just one more task to be completed, rather than a true North Star that guides the team's work.
Another problem with OKRs is that if the culture is to meet all goals 100%, teams will often choose easy goals, a practice known as sandbagging. This can lead to a lack of ambition and a failure to achieve truly impactful results. Alternatively, teams may require extensive forecasting to ensure the goal can be met, which can be time-consuming and may not be accurate.
In either case, the goals set will likely not be aspirational and setting goals will take a long time. It is important to strike the right balance between setting challenging but achievable goals, and providing the resources and support necessary to achieve them. Overall, while OKRs can be a powerful tool for achieving organizational goals, it is important to use them in a way that aligns with the team's priorities and fosters a culture of ambition and continuous improvement. By staying aware of the potential pitfalls and working to mitigate them, organizations can get the most out of their OKRs and drive real results.
The above blog post was co-authored with ChatGPT, while the image was generated with DALL•E 2. Here I will dive into my experience with the process.
AI assisted blogging
I have not been blogging much because I find it very time consuming. Hence, I wanted to test how quickly I could write a decent blog post using AI, based on some ideas I had jotted down. Specifically, I capture ideas in a text file using using my Android TODO app (Simpletask). Here is the original note:
(D) 2021-10-28 blog post on goals settings should not be about ticking a box but a core emotive exercise to drive teams +blog @computer @weekendBased on that, I logged onto ChatGPT and gave it this prompt:
Write a Hemingway style 500 word blog post based on the following bullet points:
Why are many companies adopting management by objectives, specially Objectives and Key Results, or OKRs
What are Objectives and Key results
Problems with Objectives and Key results:
1. They are mandated from above, and do not, in fact, reflect the day-to-day goals of the team. They become just one more task, not a true North Star;
2. If the culture is to meet all goals 100% teams will choose easy goals, a practice know as sandbagging. Alternatively, they will require extensive forecasts to ensure the goal can be met. In neither case is the goal likely to be aspirational. And setting goals will take a long timeThe result of the prompt is the text of the blog post above. Word for word, and without edits.
Next, I asked ChatGPT to count the words in the responses. "My previous response is 486 words." replied the AI. Then, I asked it to propose a title for the blog post. It answered: '"Navigating the Pros and Cons of OKRs: Achieving Organizational Goals with Objectives and Key Results"' (double quotes included for engineers out there). I found the title too long, so I typed: "Make the tile shorter". The answer is the title of this blog post.
Finally, I wanted a picture to enliven the blog post. Being lazy, I first tried to copy paste the whole blog post into DALL•E 2, and use it as a prompt. Maybe, because the copy paste operation resulted in truncated words, the results were not pretty.
Next, I typed the following prompt:
An illustration in the style of The Economist showing business folks dissatisfied while setting objectives and key resultsI selected the third image below to illustrate this blog post.
So, how quick is the process?
As you can see, generating 500 words of decent text from a few bullets is super fast. However, there are still some pain points in the process. For example:
Copy pasting responses from ChatGPT into my blogging platform.
Reformatting the pasted text, as it was pasted in one long string with no line breaks.
Positioning and resizing the image.
I wanted a smaller image with landscape orientation. I tried to ask DALLE for this with no luck. The default square images are not great for a blog. In the end, I went with some default layout proposed by the blogging platform.
These pain points could be avoided if the AI was fully integrated in the blogging platform (or if I was using a code based platform and calling the APIs). If a platform integrated these tools, and made it easy to switch, then I would be very tempted to switch.
Will I blog more often?
I would blog more often if I was able to remove some of the pain points above. Still, my sense is this process works best for simple blog posts, ones with a few ideas. I am not sure it would work with deeper blog posts, like the one I wrote on Automation Without Tears that go deep into a topic, and have lots of URL links. In part because writing is thinking. Often, the reason I write is precisely to clarify my ideas. I don't start with an outline. That only emerges at the end!
However, even in more complex cases, the AI could help in context to insert boiler plate text and paragraphs. For example, if you need to define AI, delegate it to the AI. Like so: "What is AI in 25 words or less". Answer: "AI is the simulation of human intelligence in machines.". That is good enough, and certainly much faster and shorter than anything I would have manged in 2 seconds. Ditto for images. For example, I can envision a version of Markdown where I could insert an image like thus:
In fact, it should not be too hard to write an extension to do all of this in my favorite text editor VS Code. Curious, I asked ChatGPT: "Write an extension for VS Code that can call ChatGPT via an API in a markdown file". Here is the answer:
Unfortunately, as a language model, I am unable to develop or create a software extension such as a VS Code extension. However, I can give you an idea of how to achieve this.
To create a VS Code extension that can call a language model API like ChatGPT, you would need to use the VS Code Extension API and a programming language like JavaScript. The extension would need to make an API call to the language model API, passing in the text from the markdown file, and then display the response from the API back in the markdown file.
You can use the `vscode` module to interact with the VS Code editor and the `node-fetch` or `request` library to make the API call to the language model API.
You would need to create a function that gets the selected text in the markdown file, or the entire markdown file if nothing is selected, and sends it as a payload in the API request. The response from the API would then be parsed, and the processed text could be inserted back into the markdown file.
Also you have to create a command that can be invoked using a keybinding or via the command palette, this command will trigger the above function and call the API.
It is a good idea to consult the VS Code Extension API documentation and the API documentation of the language model API you will be using to get more information on how to implement this extensThe future is now...
PS After publication, and reading on my phone, I noticed at least two typos in my writing, and none in the AI's. So one benefit of delegating to the AI is fewer typos, and less time correcting them.