There’s a lot of misconception about what AI is and can do. We’re here to tell the truth and cut through the clickbait. Here’s what’s up with AI today.
In a society dominated by buzzwords, “AI” is probably—at least for the moment—the buzziest of them all.
With any new trend comes a flood of clickbait, rumors, and misconceptions that get in the way of the facts. Artificial intelligence is no exception.
It’s bad enough that we have movies like Terminator 2 and The Matrix telling us that smart machines will eventually destroy human civilization as we know it. (Worse than how The Matrix sequels destroyed that franchise.)
The truth about AI might be different from what you’ve heard and read. To help sort through the noise, Daniel Canvas (100% human media director and partner) and Michael Kline (100% human web developer) put together a list of seven misconceptions about artificial intelligence (plus two bonus ones).
Let’s begin.
One fear that’s probably more pervasive and worrisome than sentient killer robots enslaving humanity is the possibility that we’ll lose something more important than our freedom: our jobs.
You don’t have to be living in today’s world to think that, either; in 19th century England, rioters protesting against the use of machinery to weave fabric and harvest crops by smashing the nefarious machines.
While it’s true that AI can automate certain tasks and perform certain operations—like pattern recognition and math—without a lot of human input, it doesn’t possess two critical functions that humans have in abundance. These include creative thinking and emotional intelligence.
Today, AI can replace repetitive motions, customer service tasks, and anything that requires decision-making based strictly on data. AI-driven machinery can put together products, direct customers to the right resources, and take your order at Taco Bell.
What AI can’t do at this present moment in time, though, is replace jobs that require empathy, human connection, strategy, flexible and subjective thinking, or abstract creativity. That means most professions are safe, even if AI can be used to help human workers in these jobs.
Your particular job might not be safe, but we’re far from the reality where human labor is obsolete as a whole.
We’d all like to believe we’re never wrong. But we’re fallible—and so is AI.
The perception that AI can never be wrong is because it can come up with a shocking amount of information in response to a query far faster than humans can. We mistakenly equate that volume and velocity with accuracy, but as anyone who has ever blindly used ChatGPT’s answers without double-checking knows, the AI does get it wrong.
The saying “Garbage in, garbage out” applies just as much to artificial intelligence as it does data science. The quality of what we put into the machine’s black box impacts the quality of what it gives us.
If AI is trained on inaccurate data, its answers will be inaccurate. For instance, ChatGPT is trained on a data set called the Common Crawl, a data set consisting of more than 240 billion web pages spanning 16 years. You can incorporate more data by adding plugins to connect to web search or even your own website analytics, but if the data you request from ChatGPT is not included in its data set, ChatGPT’s answer will most likely be wrong.
ChatGPT and other language learning machines can lie through what’s called AI hallucination—a phenomenon in which the AI system provides info that might seem plausible, but isn’t, all because the system is trying to merely replicate how humans process language. It isn’t necessarily trying to be an all-knowing giga-brain.
Not yet, at least.
In intelligence, context is everything. If we say, “The roof is on fire,” that could either mean a literal roof is literally on fire, or we’re just jamming to Rock Master Scott & the Dynamic Three.
Advancements in natural language processing still leave gaps in complex language understanding within AI, primarily because it doesn’t have the ability to understand context like we do. It tends to have more of a “black and white” approach at comprehension–it thinks like, well, a machine.
Humans, by contrast, can apply context and situational thinking to what we see and hear around us. That means we’re better at picking up on sarcasm, irony, cultural references, and subtextual meaning. (For example, what would you think if you read the sentence, “He saw her duck under a table”? You probably wouldn’t be listening for a quacking sound.
One big barrier to progressing today’s AI to a more generalized intelligence capability is giving the system the means to understand context like a human—instead of just speak like us.
When we say “AI,” we’re referring broadly to a lot of different types and degrees of artificial intelligence.
Here’s a very quick rundown of some key terms:
Not all AI systems within these categories function or are applied in the same way. New use-cases for each are being discovered constantly, and as technology evolves, the scope of each category will likely grow.
As it stands today, most AI platforms you’ll probably use—ranging from ChatGPT to Alexa—fall into the first two categories. AGI is still some ways off.
AI has changed since the release of Dr. Sbaitso from the DOS days (you young kids won’t get that reference), but one thing remains the same: that AI still lacks consciousness or self-awareness.
A platform like ChatGPT—even its most advanced versions—simply executes programmed instructions. AI has no mind, will or emotion. We can give it instructions to roleplay a particular emotion, or even an entire persona, but the output we get is still based on our instructions and the data used to train it.
We recently had a conversation with AI in which we basically created a “Choose Your Own Adventure” book, with the AI playing Doc Brown and ourselves as Marty McFly.
While the AI did have extensive knowledge about the Back to the Future trilogy and could mimic conversation like a human might, it was ultimately unable to have a real conversation–it was merely following a set of instructions based on available data.
(Your Tamagotchi pet wasn’t real, either. Sorry to break your heart.)
We’re getting into truly speculative territory here, but the idea that AI will eventually become—gasp—smarter than humans is probably more colored by how we perceive AI from movies and other forms of storytelling.
Truth is, when it comes to AI surpassing human intelligence as a whole, there is currently no concrete scientific evidence to back up that claim.
Why? Because intelligence is defined not only in the acquiring of knowledge and skills but also the application of them. As we’ve discussed before with fallibility and understanding context, AI does not have emotions, intuition or other tools that contribute to the full picture of intelligence. That means the application of what it “knows” is lacking, and it can’t create novel (new) connections and associations that we can.
Granted, in some areas, AI has already surpassed human intelligence. AI can now defeat humans in strategy games such as chess, can perform surgeries, and even operate vehicles.
In other ways, however, it’s very unlikely that artificial intelligence will ever reach a human level. Can we create an artificial intelligence that recognizes when to show kindness, or display true empathy? We can train an AI to mimic the style of a musical artist, such as the AI-generated Drake song that was nominated for a Grammy, but will we ever create an AI that can show true creativity, creating its own style?
Current AI has not scratched the surface of accomplishing emotional intelligence skills such as true creativity and true empathy, and in order to do so, we would have to teach an AI to feel. If we ever do reach that precipice, we’re introducing much deeper ethical issues than that of the impact on the marketing industry—because then we’d be creating a conscious AI and raising some very tough questions about when a machine is considered “alive.”
And we’ve all seen how those movies go…
This misconception is tricky. On the one hand, AI is not inherently biased. As we’ve said, it receives and executes based on available information and a set of instructions.
But can it exhibit bias? Well, possibly. AI can adopt the biases of the information it is given. Machine learning algorithms require human input on what to learn and how to interpret that data. So while AI is not biased, the data that it takes in absolutely is.
Of course, bias isn’t necessarily a bad thing, not if we’re trying to reflect reality as objectively as we can. (We’ll for the moment avoid the thorny argument that there is no objective reality.) But there are plenty of reasons why we’d want our AI systems to stick as closely to factual truth as possible—it’s just a matter of who defines what the truth really is.
Many of us envision an AI system as a computer that we switch on and leave to its own device—kind of like a digital Roomba.
Like pretty much anything computer-related, though, AI does require regular updates and maintenance to perform optimally (and correctly). These updates include bug fixes, security patches, and feature enhancements.
Continuous monitoring and updating of software is also needed to track and ensure accuracy over time.
If either of these are neglected, it can lead to subpar performance, inaccurate information and results, or—at worst—a system failure.
Finally, we come to what many envision as the prime goal of a super-smart AI machine: the ability to predict the future.
AI does have the ability to analyze data and trends and make predictions based on that information. Humans can do this as well. They each have their strengths
For instance, Target used big data and predictive analytics to predict when women were pregnant, before those women even told their friends, in order to determine their buying habits.
AI can use behavioral data to predict trends and common outcomes, but what AI cannot take into consideration is the unpredictable nature of real-world scenarios or events.
As one example, weather can be analyzed and predicted, but things can change in almost an instant regardless of prediction. (If you’ve ever ranted at your local meteorologist, you know what that’s like.)
Plus, the future is all about probabilities, and even extremely unlikely situations can (and will) happen. In 2015, you could’ve bet on soccer club Leicester City to win the English Premier League championship. They were 5000:1 to win, meaning that for every dollar you bet on them, you’d win $5,000. That’s a very long longshot, and no one (rightfully) gave them a chance.
They won, of course. And someone probably bought a new house as a result.
As we have mentioned, the algorithms and ability to analyze data may make AI seem like magic for predicting the future, but it is a program that is executing based on the data it is given, which is incomplete.
If you’re thinking about using ChatGPT to help you bet on the ponies or take home a lottery, save your money.
AI is moving extremely quickly. What you see today is probably lagging behind what’s going on behind the scenes in places like Microsoft, Google, OpenAI, and probably dozens of government agencies across the world.
People are excited, but they’re also worried, and both are understandable. But the reality is that the doom and gloom being tossed around is, for the moment, unwarranted.
We should look at AI as a tool that can offer tremendous advantages and capabilities, but a limited tool that is restricted to what we ourselves are capable of telling it to do. The goal is to understand AI so we can use it wisely and ethically.
Up next: we’re covering ways AI is already being used that you might not even be aware of! Don’t miss it.
It all starts with a spark, an initial plan that clearly states your goals, objectives, and how to generate real growth for your business. It all starts with a plan that lays out how we'll hit your marketing targets. We create clear strategies that guide your business towards real growth.
Anyone can run ads, but not everyone understands your audience, your strategic positioning in the market, and how to ignite an instant connection with your potential customers. This is where we come in. We launch your campaigns with speed and precision while keeping an eye on the results. Our ads grab attention, and our analytics make sure every move counts.
Your website is where visitors become customers. We build and optimize it to make sure you're seen by the right people, at the right time. We design and maintain websites that are built for conversion. We then implement long-term SEO strategies to keep your content fresh and increase your search rankings and connect with your target audience.