johnmaconline

Here is where it gets interesting and where you can fully harness AI’s current power.

Whether you are a marketer, salesperson, team leader, project manager, student, blogger, screenwriter, author, journalist, pastor, or you write the newsletter for your mom’s group, ChatGPT is your new best friend. You can and should use it to both help you create and edit. It’s like having a fabulous assistant, a group of friends you sit around with getting sloshed and talking about big ideas, and access to a focus group all in one beguiling little prompt.

For content generation, you can use ChatGPT at whatever level you want to, from generating ideas to writing whole articles.

“Write a 200 word summary article of this weekly crime data for a blurb in the police ledger of a newspaper”
“I’m writing a sermon about Psalm 23 for a liberal Presbyterian congregation that has a large income diversity. Suggest 5 themes I can explore”
“Give me 25 recipes that meet the following criteria: protein > 20g, carbs < 50g, fat < 15 g, no eggs, gluten free”
“Write an end of the year message to my team that summarizes all of our accomplishments and sets out a path for the next year”

Then, iterate in a conversational context to get to what you want.

For editing, you can ask ChatGPT specific questions.

“Edit this blurb about crime statistics so that it fits into less than 200 words”
“Here is my sermon. Suggest some improvements so that I can better relate to my economically diverse congregation”
“Here are 10 recipes. Modify them so that they are all formatted identically to the format I provided earlier”
“Edit this end of year message to my team so that it reads more succinct and ensures we keep optimism for the upcoming year”

Or you can retarget an audience.

“Here is the documentation on our software application. Create a 500 word article that summarizes it and is targeted to an audience marketing and sales people”
“Here is my article about XYZ. Rewrite it for a an audience of elementary school kids and use analogies”

This is just the tip of the iceberg. ChatGPT and LLM Ai, in general, are built for writing.

If you do any work with people outside the United States, ChatGPT is an indispensable tool for written communication.

Google Translate is a great tool, but it lacks something very important — context. You can craft some communication, formal or informal, that makes sense in English, dump it into Translate, and you’ll get back and direct translation. But sometimes, it loses meaning or even creates incorrect connotations. 

ChatGPT is much better at context and the nuances of formal and informal communication in different languages. 

Here’s an example:

“Craft a marketing brief in Italian to be emailed to our list about the launch of the software application found here: https://idex-biometrics-fuze.readthedocs-hosted.com/en/latest/index.html”

ChatGPT understands the context and nuances of marketing briefs in Italian, plus it will go figure out from the URL what that software is all about. It will create something useful and hit the mark.

Also, harkening back to Part 3, you can create a chatbot that translates languages in real-time, making it seamless to the people talking with your chatbot.

ChatGPT is bringing the world closer together and making your life easier through written translation. 

Yes, in 2024, you’ll be conversing (written) with AI quite a bit. Here’s how you can get in on the game.

You can use ChatGPT in two main ways for written conversation:

1. Real-time with chatbots
2. Generating responses for you

If you’re tech-savvy, creating your own AI-backed chatbots is within your grasp already. I won’t describe the details of how to do it here, but you can create the chatbot and train it on all of your information, then make an interface to it available to the public. 

Granted, creating chatbots appeals to a limited audience. So number two may be more interesting to you. Here’s an example of how I use it. 

I’ll get a long email thread that started way before my inclusion. Assuming no confidential material, I’ll paste the entirety of the thread into ChatGPT and ask it the following:

“Summarize this thread for me. If I am being asked to do something or respond, tell me, and then formulate a response”

Out pops a summary of the thread, an indication of whether I’m am tasked with something, and if so, an initial response. Often, that response is good enough to go as is. If the required response does need a decision from me, ChatGPT will ask me what I want the decision to be.

The bottom line: ChatGPT can help you both speed up your ability to respond and improve the responses you give.

Much of writing, if even just an email reply, requires research first. Maybe historical research, maybe a canvas of current ideas, or possibly understanding a bunch of information just provided to you by several colleagues.

How do you learn about things now?

Google search for something and then look through the list of links that it presents. This is great. Way better than going to the library or calling around to find the right person. Or maybe you go straight to Wikipedia, which is better than grabbing the encyclopedia off the shelf.

ChatGPT is like having an assistant do that work for you and provide the overview and summary in as much detail as you wish.

“Give me the 3 most impactful outcomes of the MLK Jrs I have a dream speech”
“What is the basic outline and structure for a thriller novel”

Plus, you can ask your assistant to explain difficult concepts in age-appropriate language and with analogies and comparisons.

“Use an analogy to explain quantum mechanics as if I’m 10 years old”

ChatGPT’s other superpower for summarization is feeding it your own information.

“Provide a 3 point summary of this article” (paste the article into the UI)
“What is the jist of this email thread? Am I being asked to do something?” (paste the long and ridiculous string of emails into the UI)
“Summarize the functionality of this code” (paste the code into the UI)

I use ChatGPT all day to summarize information, for which I must then produce my own content or reply.

In addition to being a software developer, I’m also a writer. So are you because everybody is to a certain degree. 

Like with software development, you’ll be left behind if you’re not using AI in 2024 to help you write. I’ve been using ChatGPT as a writing tool for many months now. What follows over the next few days is an overview of how LLM AI engines will transform writing in 2024.

ChatGPT is a general purpose Large Language Model. It’s pretty good at interpreting and creating the written word.

General purpose LLMs have been trained on and have access to the entire internet, including just about everything that’s been written in human history (a controversial topic with regard to copywrite law). Just think if you had an eidetic memory and had read everything. Literally everything. That’s ChatGPT. Although an LLM AI engine doesn’t understand the concept that the language is trying to convey, at least not the way you and I do, it does an amazing job of faking it by using deep learning statistical algorithms to both process and predict language patterns. 

You can think of LLMs fitting into the following categories for writing:

1. Summarization
2. Conversation (reply generation)
3. Language translation
4. Editing and idea and content generation

I’ll tackle each of these over the next several days.

I work in automation. I know a lot about it. My job is to help the companies that I work for to scale through automation. 

Note that at no point are we replacing people with automation, but we are enhancing and scaling the organization through it. 

There are three basic stages:

1. Personal Automation
2. Partial Automation
3. Full Automation

Personal automation is like having a basic coffee pot. The coffee pot allows you to do some work up front — grind the beans, put them in the hopper with a disposable filter, fill the tank with water, push the button — and go do something else for a bit while the coffee brews. Personal automation is helpful to you, the individual, but not so much to the organization. It doesn’t move the needle.

Partial automation starts to gain benefits for the organization. Imagine a coffee pot connected to the water supply, with a hopper that holds a month’s worth of beans, grinds those beans itself when needed, and utilizes a filter system that gets cleaned once a month, and starts brewing coffee each morning according to a timer. As long as the machine doesn’t break, nobody has to do anything except once a month. That’s helpful to both the people and the organization.

Full automation is the holy grail. This is where automation can scale your organization. Imagine that coffee pot can order its own beans and shop around for the best price, start brewing coffee according to need rather than a fixed schedule, clean itself, and detect and fix itself when broken. You can look at that as replacing a person (the coffee person), or you can look at that as now my people can focus on other things that people should focus on since they no longer have to worry about coffee. 

Today, for software developers, AI is at the personal automation stage. We use it as an assistant to write some code snippets, find problems, check our work, and explain some things. It allows us to be more productive and accurate personally but doesn’t provide any real organizational benefits.

But we’re on our way to partial automation. 

Just a few weeks ago, I wrote to my boss that “scope” was one of the limitations of the interactive UI by limiting practical use to 300-400 lines of code at a time. 

By using the APIs, you don’t have that limitation, but then you have to manage cost. The interactive UI costs a flat fee of $20 a month. But to use the APIs, you get charged by the amount of data in and out. Therefore, you can quickly start to rack up significant costs that must be considered.

Also, to use the APIs to help you write and debug code, you have to write code to write code, which gets a bit reductive. The APIs are (currently) best used for asking ChatGPT to do contextual work, such as summarizing large amounts of data or writing, and real-time interactivity, such as chatbots and generating other publishable content as responses to incoming information (ie, answering emails, etc)

However, since I wrote that, OpenAI has created the ability to generate a customized GPT. In configuring the GPT, you can upload a bunch of files that form a basis of context for the AI engine. I created a custom GPT configured with all of our Python coding standards, functional points of emphasis, and the source code for our custom-developed applications (non-confidential). Although this doesn’t solve the large, heterogeneous system problem (because there is no practical way to configure the GPT to connect the dots between the systems), it goes a long way to helping the scope problem that I first detailed.

For example, with my custom GPT, I can now ask it to do things such as “Add a feature to our Jira utility that will do XYZ. Make this function available via the –xyz argument.” Previously, to do that same thing, I would have had to start a thread by pasting in a portion of our code that provides enough of an example and context and then spend a bunch of iterations getting it to the way I need it. Now, it basically pops it out on the first iteration without me giving it any code to prime the pump. 

Part 3 tomorrow. 

I’m starting with software development because this is my wheelhouse.

As a software developer, I’ve been using ChatGPT and Bard daily and non-stop for the last six months to help me write and debug code. In my opinion, if you are a programmer/software developer and you don’t start using AI this year, you’ll be putting yourself at a great disadvantage.

It’s just too good at helping you not to use it. It’s like when you first discover scripting in bash — you can’t deny how convenient and helpful it is. Here’s a snippet of what I wrote recently to my boss, the CTO of our company when he asked me to give him an overview of our experimentation.

Pros

• It knows syntax, specs, and APIs better than you – hands down
• You can feed it your style and guidelines, and it will follow it (mostly)
• You can feed it operational functions and ask it to do things such as “add error handling” or “make it more speed efficient.”
• It’s good at debugging.

Cons

• Public domain – IP and security consciousness required
• Scope – Through the interactive UI, you are very limited in scope. 300-400 lines of code are the practical maximum. Therefore, it’s not all that helpful at the application or system level. 
• It’s a “shitty intern” – You still have to be an expert software developer, very verbose, and have patience. Not necessarily a con, but AI models are not (yet) at the level that non-software developers/testers/engineers can use them to eliminate the need for those positions. Basically, you can’t replace the people (yet) and save that cost.
• It’s not foolproof – We, the humans, must still be the adults in the room. In many ways, our jobs as verifiers become even more important. There’s a trust issue to manage. 

Part 2 tomorrow.

If you thought 2023 was the year of AI, think again!

AI will dominate the headlines even more in 2024 than it did in 2023. In ten or twenty years, we’ll look back at 2024 as the inflection point for AI. 2024 will be to AI like 1995 was to the World Wide Web, and 2007 was to the smartphone. 

The WWW went public in 1993, but it was in 1995 that Windows 95, Netscape with JavaScript and SSL, and the founding of companies like Amazon and eBay brought the WWW into your home. Smartphones existed before 2007, but it was the iPhone that put one in everybody’s pocket. 

Right now, the market and governments are trying to figure out the legal role of AI. Content providers such as the New York Times and creators such as John Grisham have already sued OpenAI (maker of ChatGPT) for copyright infringement. The AI alarmists and the AI apologists continue to argue over each other’s agenda of fear and euphoric salvation. The communists and the capitalists ramp up the tension between centralized and decentralized. 

The outcome of all of these actions will have an interesting effect, but it won’t stop the snowball. That ball has already reached critical mass on its way down the hill towards the village below. 

Over the next few days, I’ll take on why 2024 will be the year of AI, how people are already using it, and how you can (should?) bring it into your own world.