If you have not heard of ChatGPT yet, I highly recommend you check it out. And if you want me to tell you, how I about I just show you? Hey GPT, how would you explain yourself to a consulting agency in 4 sentences, focusing on how you could add value to technical and business consultants?
ME: Hey GPT, how would you explain yourself to a consulting agency in 4 sentences, focusing on how you could add value to technical and business consultants?
GPT: Hello! I am a large language model trained by OpenAI. I am able to generate human-like text based on a given prompt or context. I can assist technical and business consultants by providing accurate and relevant information on a wide range of topics, as well as by generating reports, summaries, and other written materials. Additionally, I can help with data analysis and research, allowing consultants to save time and focus on their core competencies. Overall, I can add value to consulting agencies by providing high-quality, reliable support for their work.
Well, maybe that was five sentences, but that’s pretty awesome. For this blog, I will tell you why I think ChatGPT is the most useful tool in the world.
GPT Would Pass Your Coding Interview
Alright GPT, that’s pretty cool, but what about programming? What specifically could you do to help a developer who knows they need to loop over a 2D array named positions, access the variable amount on each item then reorder the array in descending order?
GPT starts off strong, first giving us high-level information.
Code response from GPT detailing high level 2D forloop syntax in javascript
And it continues, outputting what could be put straight into a project with only a little modification:
ChatGPT’s second part of the coding response
Even crazier, it was able to decide that because I used the word ‘amount’, it would swap them via subtraction, rather than memory swapping. Of course, maybe we wanted it to swap the object… Well hey let’s tell it!
This is one reason GPT is the most useful tool in the world — and I’ll go on to explain use cases for non-programmers too. Think about all the different languages you might need to code in over the course of a week, a month. How many times do you know conceptually what should happen, but the precise syntax escapes you? Well not any longer.
GPT Must Have Attended SUGCON
Here’s an example of something a colleague shared with me:
GPT’s Sitecore powershell answer on getting all references to an item in all languages
GPT remembers context too, meaning that you can build off previous responses and have it put together more and more of an answer, without abruptly forgetting what it was you asked before. I’ve googled “how to kill child zombie” before remembering to specify I was programming in C, but in GPT if the last question I asked was about C, it would automatically assume my intent, which is really cool.
GPT For MVP?
You can do far more than program with it though. For example, Sitecore’s new headless architecture is gaining traction, what are some of its benefits?
Sitecore is a popular content management system (CMS) that allows users to create and manage digital content. “Headless” Sitecore refers to the ability to use the CMS as a back-end platform for delivering content to various front-end applications, such as websites or mobile apps. This allows for greater flexibility and control over the way content is presented to users.
An example of where headless Sitecore might be useful is in the case of a company that has both a website and a mobile app, and wants to use the same back-end CMS for managing the content on both platforms. By using headless Sitecore, the company can use the CMS to create and manage their content, and then use APIs to deliver that content to the website and mobile app in the appropriate format. This can save time and resources, as the company only has to manage their content in one place.
Amazing. What might have required multiple Google searches and speed reading is instead parsed into an easily digestible paragraph in seconds. Now, it’s not without its limits, so let’s briefly mention those.
Limitations
The bot trained on data that ends somewhere around 2021-2022, and so asking it to synthesize responses on topics after that does not work well. Additionally, directly querying it about current events will often trigger it to tell you this:
I’m sorry, but I am not able to browse the internet and therefore I am not able to provide you with the current version of React. As a large language model trained by OpenAI, I am a virtual assistant and do not have access to the latest information and updates on specific technologies. I can provide general information and answer questions based on my training, but I am not able to provide up-to-date information on specific technologies or applications.
The bot is also able to get things wrong. When asked it to cite some sources for a section of writing, and it gave him three sources. Two were great, while the third listed a book that didn’t exist. For coding, the right and wrong of it is pressing compile, but for free response be careful. When it declares that mass and acceleration are unrelated, it could be wrong.
There are also other concerns to consider, expertly detailed out by Brandon Luhring.
Closing
Even with limitations, I think ChatGPT could be the most useful tool in the world. It can recall context, synthesize across thousands of topics, and remember detailed specifics. Just imagine a future where a bot like this can study your work and provide increasingly accurate suggestions. Consultants have always been about solving a problem, and AI might finally let us focus solely on that.
Bonus Round
Can GPT be a SCRUM master?
Hey Drew! I’ve been using chatGPT over the last couple of weeks for questions I would have previously done through regular searches. Getting a simple explanation of say a property or method you’re not familiar with is a great way to get into practice, as this is more like asking a person for just what you need clarified, and of course any example how it would be used.
I’ve found it very helpful when doing some kind of training and there is something not fully explained and you just want to fill that gap so you can continue on.
You give a good example of the kinds of responses to expect, I made the mistake of asking it about housing prices and that didn’t work at all.
Thanks Ed! I’ve found it useful in the same way, plus the ability for it to generate templating in most languages with caveats that I provide. Just today I asked it to create “modern hover states for a link” and gave it a few colors. The result was amazing, I pasted it straight in!