AI is dumb. We said it. We meant it.
Okay, that deserves some explanation. First off, let us say that we do not hate AI. We actually find it quite useful. We know everyone is super excited about AI and rightly so. It’s going to change everything.
It’s still in the early stages, though. If you’ve used it at all, you probably recognize that it’s going through some growing pains. While it’s wonderful for some tasks, it’s not so great for others.
Artificial intelligence is not actually “intelligent,” at least not yet.
ChatGPT pros and cons
Take ChatGPT for example. It’s a fantastic research assistant. You can ask it anything, and it will scour its data source for answers. Just be sure to fact-check, because there is a lot of incorrect information out there and it can’t tell the difference.
ChatGPT is an excellent tool for gathering ideas, especially when you’re faced with writer’s block. It can put together an outline for an essay, summarize complex ideas, generate headlines, and that’s just the tip of the iceberg.
But try to use its output verbatim, eh…not so much. We find that it writes like a thesaurus, overusing buzzwords and phrases like “elevate,” “unlock,” delve,” “uplift, “dive into the world of,” “unleash your creativity,” and “in today’s ever evolving landscape.”
In our opinion, you can’t just copy and paste what ChatGPT spits out, at least not if you want original, well-written prose that doesn’t resemble everything else on the internet. It still requires a human’s critical thinking skills and creativity to edit and refine the copy.
Here’s a little tip: Tell ChatGPT not to use poetic language, and you’ll get much more human-sounding response.
Why is AI dumb?
ChatGPT 3.5 doesn’t know Queen Elizabeth is dead. It doesn’t know because it’s not (as of right now) using real-time data.
Despite having “intelligence” in its name, AI doesn’t actually think and it’s limited by the data it can access. It’s incapable of synthesizing real-world experiences and observations, fostering creativity, developing original thoughts, and has no self-awareness.
Instead, it uses machine logic and derivative assumption to take human-created information, search it, and spit back answers. It’s essentially a parrot that uses logic, math, and advanced plagiarism to repeat our knowledge back to us.
Think of it this way: Information is stored in eggs. ChatGPT searches for the eggs that match your query, then whips them up and serves them back to you sunny side up, over-easy, scrambled, poached, or in an omelet depending on your preference. It’s not creating new eggs, original recipes, or devising better ways to cook using real-world experiences. It’s simply gathering information from a database of existing human thought.
The fact that AI draws from other people’s knowledge and creativity leads to a whole host of ethical and copyright issues that experts grapple with. But that is a topic for another day.
AI’s drawbacks
You’ve probably seen examples of generative AI gone wrong. While it can be useful for creating amazing landscapes and objects, for some reason it has a really hard time with humans. It makes us look alien, with strangely-angled or extra limbs, fuzzy hair, extra fingers, and other distorted features.
Full disclosure: We used Canva’s AI tools to create the computer images in this blog, so it does have its uses.
To get good results from generative AI requires skillful, detailed querying and refinement over time. AI will continue to improve, but for now it can’t fully replace the creativity and originality of the human mind. Yet.
The future of AI
The next step in the evolution of AI is AGI, artificial general intelligence, a system that could imitate human thought using logic and reasoning. While AGI is possible in theory, experts say we’re not even close.
For one thing, humans use more than just the written word to gather information. We learn from the full scope of our experiences, through observation and interaction with the real world around us, something machines obviously cannot do.
Some experts even say it’s a misnomer to refer to these tools as “intelligence,” since AI is simply repeating our own knowledge back to us. Essentially, there’s no brain behind the computer and there may never be.
So don’t worry just yet, John Connor.
