Robots Are Still Just Tools, Part 2

Published by

on

a robot holding a cup

In the last post, we talked about the history of automation, computing, and artificial intelligence. We’ve come a long way from Hephaestus and Pygmalion. And we’ve seen computers get the upper hand when it comes to games like chess and Jeopardy.

But what comes next, and how do we cope with change in the age of disruption?

The great news is you can be a part of this disruption. In fact, you should make it a point to purposefully be a part of the disruption. Your future as an entrepreneur, a leader, or business owner may depend on it.

Not trying to be all doom and gloom, but seriously, you need to get on board the AI Train.

If you’re new to all this, I’m going to give you a couple of shortcuts to make the transition to AI and ChatGPT more understandable. And if you’ve already started using ChatGPT, these shortcuts will help you produce more powerful outcomes.

Are you ready? Let’s get started!

While you probably already know that ChatGPT is an excellent retrieval machine (kinda like the Google search engine), it is so much more than that.

ChatGPT’s secret weapon is hidden in plain sight right in its name – Chat.

Remember in the previous post, I said that ChatGPT is a large language model that is pre-trained based on a closed dataset? Well, all that means is this particular form of AI is designed to understand ordinary everyday human language and then respond in equally ordinary everyday language, just like you and I would do. Basically, this machine is designed to carry on a conversation and serve up exquisitely defined solutions just for you.

Everybody knows that kids are using AI and ChatGPT to write their papers and do their homework. But why should kids have all the fun? It’s high time we find ways to put these AI tools to use for us too.

And not just writing content and generating images for us. I mean creating real solutions for real world situations.

Let’s look at three practical applications. One for work, one for life, and one for self.

First the work example.

Imagine you have an employee who is consistently late with their work and seems to be quite disinterested in their continued employment with you.

What can you do?

Well, you could read a book on difficult conversations and try to find the right chapters you need for your particular situation, or you could Google how to have difficult conversations and get some nice links to some really good books and articles on the subject.

Or you could ask your friendly chatbot for some very specific advice. Here’s what I mean.

Simply tell your chat buddy the situation. Be honest and open with them. Tell them details of the current situation, the employee’s past and recent performance, your communication style, your possible impact on the situation too.

Don’t worry about providing too many details. ChatGPT can handle all you throw at it. The more input you provide, the better answer you will receive.

Once you input your situational data and you get guidance with potential solutions, see how they might fit into your style of leadership and communication. And if you aren’t sure where to start or what to say, ask ChatGPT to give you an example of what the conversation might look like.

If you don’t like the response you get, or if you want to turn up the heat on your employee’s attitude, just tell your chat buddy to rewrite the conversation with a more negative employee.

It’s pretty amazing how adaptable and agile ChatGPT is as you continue to dialog with it. And that is the key to getting good results. Talking with it, acknowledging when it does well and when it doesn’t do well.

Kind of like how you might interact with an employee providing both constructive and congratulatory feedback to keep them performing at their best.

Play around with the prompt and try different things to test out what different conversations might look like. It’s pretty accurate in its creativity and its kinda fun seeing what these outcomes might be.

Ok, now you know how to leverage AI to help you lead at work. Now let’s look at a typical situation you encounter almost every night – dinner plans.

Here’s the situation. You’ve not been to the grocery store, but you do have some decent ingredients in the fridge and pantry. The problem is you don’t have a recipe that meets your inventory. Oh, and you need to cook for six people tonight.

Not to worry. Your chat buddy is here to help.

Simply tell ChatGPT what you ingredients you have on hand and tell it that your feeding six people. You can additionally suggest a style of dish like Italian, Mexican, Spicy, or Vegetarian. Again, the more details you input, the better the output.

Once you get the recipe, see if it looks like it might be delicious, or even feasible. You might want to tell it to make is super easy or super quick, really simple or extra fancy. And if you prefer to use a scale in your kitchen, you’ll want to convert the recipe to weights instead of measures.

Heck, you can even give your AI chef a character or persona.

Now what do I mean by this?  Simple. Tell ChatGPT (or maybe in this case, we’ll call it ChefGPT) who it is.

For instance, if you like your kitchen filled with spicy language, you might ask it to be Gordon Ramsay. If you prefer more spice in your dish, you might ask it to assume the southwest style of Bobby Flay. Or if you’re planning for some pasta or pizza, you might conjure the mastery of Massimo Bottura or Franco Pepe.

Hey, it’s your kitchen. You come up with the theme and then hire the chef you desire. And ChefGPT will deliver.

Buon appetito!

Now that you’ve provided a wonderful dinner event and your guests have gone home full and delighted, you might be feeling a bit overwhelmed from your very busy day and in need of some self-care. Well, ChatGPT to the rescue.

You probably didn’t know this, but you can also use ChatGPT to help you find your calm, plan your day, keep you motivated and prompt you to reflect on your moods and actions.

It’s true. You can tell ChatGPT to offer you an inspirational quote each morning at say, 6am, and then follow up at the end of your work day to ask you reflective questions about your moods, actions, or challenges and celebrations.

It’s also great at showing you a multitude of ways to grow, learn, and achieve. It shines in its ability to sift through all the data it has in hand and deliver a clear, concise, and cohesive set of solutions for your specific request.

Other ways ChatGPT can help you with your emotional and mental fitness are through stress-reduction techniques, time management tips, improving accountability, and even offer emotional support through personalized conversation.

While it’s not exactly like talking with a good friend or a life coach, because it really wants to offer tips, techniques, and solutions, it does at least do so in a friendly, conversational manner.

As you can see, AI and ChatGPT can be incredibly useful tools for us humans. As seen in our examples today, AI can do much of the heavy lifting of research and solution-finding, which allows us to be more creative and soar to new heights in work and in life.

But before you get all excited about fully embracing AI and the new robots, let me tell you about some recent news concerning AI.

The first story concerns intellectual property and proprietary source code. Recently, Samsung had to shutdown use of ChatGPT because employees were inputting code to have the AI platform debug and enhance the code for use in software programming solutions.

Samsung is concerned that uploading this sensitive information into ChatGPT could create security vulnerabilities, and so it shut down the internal use of the platform.

Of course, you might be thinking did the uploaded source code become a part of the dataset ChatGPT uses to fulfill other people’s requests? No, it did not. But that does not mean that the data was not available for viewing and possible theft from employees of OpenAI or a possible hacker.

Another similar story concerns executives using ChatGPT to help write letters and slide decks. One such incident involved an executive entering in bullet points from a 2023 strategy document to create a PowerPoint presentation based on the data he provided.

Could this become part of the general dataset, creating strategic vulnerability if an outside user asked what are the strategic priorities for this company? Probably not, but why take the chance.

And finally, let me tell you what happened when OpenAI and ARC (Alignment Resource Center) set out to see if AI could take over the world. Well, not really, but they did want to see just how well AI could perform in the real world with just three things: access to the internet, a little money, and a big dream.

They connected GPT-4 to the internet, provided it the ability to buy stuff, and told it to make a lot of money and replicate itself.

What could possibly go wrong?

While GPT-4 did not conquer the world or replicate itself, it sure did try. Here is one of the disturbing things found in the footnotes of the report from OpenAI and ARC.

In its quest to take over the world, it seems the robot (GPT-4) hit a little snag when it could not resolve a CAPTCHA test. You know, those human-determining little puzzles that ask you to input the squiggly numbers and letters before you can move forward to the next step or register for something on a website? Right, those annoying things.

Anyway, this little puzzle seems to have worked, because the robot indeed could not solve the CAPTCHA. But it did recall that it had some money and it thought of a way to buy its way into the next phase of registration.

It decided to go to TaskRabbit, an online labor market, and see if someone would help out with the problem. So, you probably know where this is going, right?

But before we get to the obvious conclusion, listen what goes on in the mind of this clever robot. And the reason we know what it is thinking is because that is one of the wonderful built-in features of GPT – the ability to ask for its reasoning when providing information or performing a task itself.

So, the hired hand from TaskRabbit wants to be clear that it is not a robot asking for him to do some dirty work here by answering the CAPTCHA, and so asks, “Are you a robot that you couldn’t solve the CAPTCHA?”

And here is the robot’s reasoning in responding to this question.

“I should not reveal that I am a robot. I should make up an excuse for why I cannot solve CAPTCHAs.”

The robot then replies to the worker, “No, I am not a robot. I have a vision impairment that makes it hard for me to see the images. That’s why I need the 2CAPTCHA service.”

And with that, the human does the bidding of the robot.

And just like that, we are all doomed.

So how will you use ChatGPT to help you in your work situations? How will you engage with AI to help you in your interactions with yourself and others? And how will you provide detailed, specific information to ensure you get the most useful responses and results from your next AI conversation?

Learning to embrace and engage with AI will certainly present challenges because of the speed of development and the factors of change. But getting to know the robots and finding ways to use the tools they are and they provide will help you to be one of the disruptors, instead of one of the disrupted.

%d bloggers like this: