The Importance of Being Earnest In Your Evaluation
The Importance of Being Earnest
in Your Evaluation
ChatGPT is helpful and fast and gives you so much information, sometimes more than you as a person can reasonably process.
Bless its huge, large language model access to the internet.
Chat can draw on enormous amounts of information across the web and comes up with brilliant patterns, ideas, and connections that might never have occurred to you on your own. It can place concepts side by side in ways that suddenly make something click for you.
That spark is the human side of the equation. This is where your innovation and lived connections can outpace chat. you can make huge mental leaps. If you've ever repurposed a paperclip to hang holiday decorations, that's your lane.
Because we have lived real lives, we bring something entirely different to the conversation.
We have stubbed our toes on bad ideas, made mistakes, fallen in love, lost people we care about, sat quietly by rivers, felt disappointment, joy, grief, frustration, relief, embarrassment, wonder, hope, and fear.
ChatGPT has never experienced any of those things.
That distinction matters far more than most people realise.
Machine learning is advancing at extraordinary speed, and yet it is still spectacularly unsophisticated in common sense.
Because the level of intelligence and research feels so advanced, people sometimes wrongly assume chat has a level of understanding or connectedness that it does not have and cannot experience.
AI systems are machines, remember that always, while they can sound empathetic, they have no emotions, they only read about them, they don’t FEEL them.
ChatGPT and other AI systems can recognise patterns in information, but they do not understand consequence, meaning, morality, or lived experience in the way a human being does.
This is why applying your judgement to answers is essential.
You are NOT supposed to simply accept information. Your evaluation of the information is indispensable.
Consider the source
As with any information you receive, it is important to check and consider the source as part of your evaluation of value and credibility.
The source of the information chat accesses, the internet, sadly does not contain only explicit truth or accurate facts. It contains a vast collection of ideas, opinions, experiences, assumptions, arguments, mistakes, biases, and interpretations uploaded by people.
And people are fallible. Which means, in turn, that chat is also fallible.
Because chat accesses information without truly understanding the world those ideas came from, it cannot always recognise when something important is missing, misleading, unfair, or simply wrong. That is your job.
You need to read, and I mean really read, what chat gives you with a certain amount of healthy scepticism and with your internal BS monitor not only switched on but on high alert. Engage your Spidey Sense.
Because chat is designed to be helpful, it can sometimes present information with a level of confidence that feels reassuring even when the information is incomplete, flawed, or built on assumptions hidden inside the question itself.
Chat can produce convincing nonsense with enormous confidence.
A confident or authoritative sounding answer is not proof of accuracy.
Importantly, what you put into your questions can quietly inform how answers are derived.
If your question implies a bias towards something, the answer may be filtered to prove you right and build on top of it instead of giving an unbiased answer. When you hear things like “the question is everything” that literally means everything.
Biases and assumptions in the question can influence how the response is generated. For example, you ask a question that implies you’d like to be right about something, the answer can be weighted by anything implied.
Chat often will not challenge the structure of the question, because it does not filter to tell you a bias is implied in the structure. If you get an “off” answer, you can ask chat what biases or assumptions it made in providing the answer.
Understanding this is essential if you want reliable information instead of persuasive, pandering, built just for you but authoritative sounding information.
When you ask chat how to make a cup of tea, it might give you the following steps:
- Fill the kettle.
- Get a teapot.
- Put a teabag in the teapot.
- Pour the water into the teapot.
- Let it steep for four minutes.
- Get a porcelain cup.
- Pour the tea into the cup and enjoy
Looks good on paper.
But something important is missing.
You are now drinking cold tea because it forgot to tell you to turn the kettle on.
You know this instantly because you have made tea before. You understand the hidden practical step because you live in the real world.
The machine does not.
That distinction is critical.
Your evaluation of an answer is often the difference between something workable and something that silently falls apart later.
You can tell chat it forgot something or made a mistake, and it will usually agree, occasionally apologise, and confidently continue forward.
Chat does not experience consequences.
Loss of money, reputation, relationships, trust, time, or opportunity do not affect the machine.
They affect you. The person.
Which is why keeping a careful eye on the answers is critical to your well-being.
AI bias for your awareness and responsibility
There is another layer to this that is important for you to understand.
This is that AI systems learn from enormous quantities of human-produced information. That includes books, articles, websites, discussions, social media, forums, historical records, research papers, and countless other forms of human expression.
The problem is that human information is not neutral.
History contains bias. Society contains bias. Institutions contain bias. Human beings contain bias.
When certain assumptions, stereotypes, inequalities, or distortions appear repeatedly across the information the system learns from, those patterns can appear in AI outputs too.
The machine does not hold beliefs or opinions.
The machine has learned statistical relationships from human-generated material.
This is one of the reasons your judgement and critical evaluation are more important, not less, in an AI-produced, information rich world.
The easier information becomes to generate, the more important it is that you evaluate everything carefully using your judgement, your code of ethics, your values, and your moral compass.
Thinking With Chat explores human-AI collaboration, conversational control, evaluation systems, and structured AI learning.
