Debunking 9 Common AI Myths in 2024

If you've been holding off on exploring how artificial intelligence (AI) can benefit your workflow due to concerns about how "sentient" or mysterious the technology seems, we're here to put your mind more at ease.

At Upwork, we've been exploring generative AI and how it can help our customers as well as our own operations. Let's break down the most common AI myths and uncover why they aren't true.

Understanding the origins of common AI myths

A lot of common myths about AI—it's all-knowing! It's going to take over the world! It hates humans!—stem from popular culture.

From the 1927 film “Metropolis,” which involves the creation of robots with human likenesses, to supercomputer HAL in 1968’s “2001: A Space Odyssey” and “Blade Runner” in the 1980s, movies have long used duplicitous, uncontrollable AI as a plot device.

The same goes for books—stories by writers like Isaac Asimov and Arthur C. Clarke talked about AI that goes off the rails.

Granted, not all depictions of humanoid, sentient AI have been alarming—Lieutenant Commander Data from “Star Trek: The Next Generation” is depicted as a wise, sensitive, and eventually emotional AI that is a valued part of the Enterprise crew.

Scary or not, though, all these depictions of AI are far from reality. The AI that we find powering tools like Claude can't think, feel, or emote like a human. They can’t take over the world, understand our emotions, or replace human cognition.

But since many of us have been exposed to pop culture ideas about AI for decades—and have only interacted with AI since ChatGPT launched in 2022, it's understandable that sudden enthusiasm and acceptance of the technology can feel jarring.

Why debunking AI myths is crucial

It's important that we understand the myths behind AI now, right as the technology is starting to become more important to business—and life in general.

There are countless ways we can use AI in our world, including:

• Improving personalization and customer experiences
• Automating routine processes in healthcare, manufacturing, and business applications
• Speeding up the rate at which we can review and glean insights from data
• Developing customized tutoring plans for students

AI has the potential to create new classes of jobs, too, such as AI trainers and AI content assistants.

By distinguishing facts from fiction, we can better understand how generative AI and other AI systems are able to help us work, live, and play in new ways.

9 AI myths, debunked

These are eight of the most common AI myths you may have encountered—and wondered about yourself.

1. AI can think, feel, or talk like a human
2. AI understands content like humans do
3. AI technologies can learn on their own
4. AI can surpass human intelligence in all areas
5. AI will create job losses and widespread unemployment
6. AI is only useful in tech industries
7. AI is only for use by technically skilled people
8. AI is unbiased
9. Adding AI to your business guarantees higher productivity
   ‍

Myth 1: AI can think, feel, or talk like a human

No matter what sci fi books and movies have portrayed in the past, there’s no humanity at all in the AI platforms that we use today. No matter how realistic a conversation with ChatGPT feels, it’s just data-powered algorithms at work.

The way AI works is based on the data that humans have fed into it. For example, ChatGPT’s creators used content available online up until late 2021 to train their AI. They loaded all this data into a machine learning program, then created algorithms to process the data, recognize human speech patterns, and return results based on probability.

This training is why conversational AI platforms say they do or don’t understand something, issue “apologies,” and give the appearance of empathizing. The algorithms have been trained to do so—it’s what a human would be likely to say based on probability.

Myth 2: AI understands content like humans do

AI engineers use techniques called natural language processing (NLP) and semantic understanding. AI algorithms trained with NLP and semantic techniques are able to process and interpret results from normal human speech. This can include colloquial or slang language.

NLP helps AI algorithms sound more natural when returning results to human users. It’s all based on data and probability, though—there’s no reading or comprehension taking place.

The MIT-IBM Watson AI Lab and Harvard NLP created a tool called the Giant Language model Test Room (GLTR) that can show you how likely it is that AI will choose a particular word based on what it’s been told about human speech.

I asked Claude to describe a farm, and then fed its results into GLTR. Every green word that you see below is the word, or root of a word, that’s most likely to appear in that specific spot. (Purple, yellow, and red represent lower probabilities.)

This likelihood is calculated by the AI based on the vast amounts of online content used during training.

Just look at all that green!

We can now see that Claude’s response wasn’t actually intelligent or creative. It’s just … math.

Myth 3: AI technologies can learn on their own

When you interact with a tool like ChatGPT, you’re using a type of AI known as narrow AI.

Narrow AI algorithms cannot learn on their own like a human brain. The algorithm may continue to process new data that’s fed into its interfaces, but what it can interpret and do is limited by its fixed programming.

For these AI models to “evolve,” humans have to code and train a new version of the algorithm. When ChatGPT-3 turned into ChatGPT-4, it was entirely due to the work of human engineers and data scientists.

Myth 4: AI can surpass human intelligence in all areas

Human intelligence isn’t only about absorbing data. Our abilities to experience intuition and common sense are big parts of how we interact with the world around us.

Because narrow AI tools are limited by their programming, they’re never going to develop or surpass human intelligence.

Yes, narrow AI may be able to process datasets faster than a human brain, but that doesn’t mean it’s able to make decisions about that data better than a human.

If you spend any time reading about AI topics, you may come across the idea of artificial general intelligence (AGI)—it’s sometimes called “general AI” or “strong AI,” too. You could imagine that AGI might gain an understanding of what it’s like to be a human, rather than rely on algorithms to dictate its responses.

The thing is, though, AGI isn’t real. It doesn’t exist outside of science fiction.

Might AGI ever exist? Possibly, but it’s not something we’re close to right now. In a statement about the future of AGI, OpenAI says that progress toward such an AI is something that will happen “over a long period of time.”

Myth 5: AI will create job losses and widespread unemployment

It’s true that AI may lead to the loss of some jobs as we know them today. But further adoption of AI is going to create new jobs at the same time; we’re already seeing that begin to happen.

Upwork research shows that 39% of companies are currently mandating the use of AI tools at work, and another 46% of companies encourage the use of AI.

This widespread use of AI is already creating new roles in the workplace, such as:

• AI artists
• AI researchers
• AI chatbot developers
• AI writers

The World Economic Forum compares this shift to a time back in the 1700s, when an actual human job was to go around to houses and wake people up. Those folks were put out of work when the mechanical alarm clock came into existence. And this has happened throughout history—some jobs end, and others begin.

Myth 6: AI is only useful in tech industries

AI is useful in lots of different industries—as well as in your personal life. Compose a simple question in ChatGPT and review the results:

The real-world applications of AI extend well beyond technical and creative work and can include:

• Improving healthcare data analysis, which can lead to developments in medical diagnoses and drug development
• Increasing the speed at which financial transactions and financing applications are processed
• Providing travelers with round-the-clock resources and answers to questions
• Helping homeowners better manage the efficiency of their heating and cooling systems
• Pinpointing the need for maintenance in complex mechanical systems
• Creating education-focused assistants that can help students better understand new material

These are a fraction of the potential uses for AI in business and life—and as new industries and roles develop as a result of this technology, its potential business applications will continue to expand.

Myth 7: AI is only for use by technically skilled people

While AI was once the purview of very technically skilled people, advancements in generative AI have made AI accessible to anyone with an internet connection.

You can use normal, natural language to write AI prompts for tools like ChatGPT, Midjourney, Jasper, and more.

While you can make intentional tweaks to your prompts that will influence the AI’s output, ultimately it’s very much like having a conversation—just remember that you’re chatting with a predictive algorithm and not a person.

Myth 8: AI is unbiased

Humans have created and trained AI—so it stands to reason that AI can wind up mimicking our biases.

According to researchers at the U.S. National Institute of Standards and Technology (NIST), the existence of bias in AI can start at the point of development.

Humans have biases—personal and systemic. These biases may be expressed in movies, books, TV shows, social media posts, online comments, blogs, and other content that’s used to train AI algorithms.

And while AI developers may attempt to create as unbiased an AI as possible, their own biases may prevent them from catching all bias present in training materials. Not to mention the sheer volume of data they’d have to sift through to do so.

If bias makes its way into an AI tool, it’s possible that the AI’s outputs would skew toward or against a particular group of people. This becomes increasingly problematic if the AI is used to make decisions that impact people’s health, lives, finances, well-being, and more.

The National Institutes of Health, IBM, governments, and other groups are considering ways to potentially reduce bias in AI. Part of this can involve AI bias audits that analyze a machine learning algorithm’s outputs.

As individuals, we can all take steps to help combat bias in AI, too. While we may not be able to control the outputs produced by an AI tool, we can keep an eye out for misinformation, spoofs, or fakes—and then refuse to spread it further.

Myth 9: Adding AI to your business guarantees higher productivity

A disconnect is happening right now between leaders who anticipate AI adoption making their businesses more productive and employees struggling to keep up with increasing workloads.

AI can boost a company’s output and productivity, but simply adopting the tool isn’t always enough. If systems and processes are already broken or experiencing friction, the addition of a new AI tool may exacerbate those issues, not fix them.

This productivity paradox has happened before—when industries began to adopt technologies (like the computer!) that are now part of our everyday workflow. AI can reach that point too, but it’ll take careful planning, systems evaluation, training, and mindful integration along the way.

Overcoming misconceptions about AI at work

If you start to play around with generative AI tools, you might be surprised to find they’re a lot less intimidating or alarming than you first thought.

And for as helpful as they may be, these tools can return nonsensical, wrong, and absurd responses, too. Once you’ve interacted with one for a while, it’s easier to see why human decision-making is a necessary part of using AI.

By being open to these new technologies, though, and taking the time to learn how to use them, you’ll be keeping both yourself and your company ready to leverage useful AI applications now and in the future. This know-how can help you keep pace with competitors, be more productive, and even explore new business opportunities.

Resources to learn more about using AI at work

If you or your team would like to learn more about how to best use generative AI and automation in your work, check out these resources for reliable information:

• Upwork’s AI services resource center
• IBM’s AI Essentials YouTube playlist
• MIT News articles about AI
• Wired’s AI updates
• Harvard Business Review’s articles about AI and machine learning

Working with AI on Upwork

You can also get help with understanding and implementing AI right here on Upwork. Whether you want to take AI and machine learning courses, work with a generative AI pro, or even begin finding AI jobs that you can do as an independent professional, it’s all here. All you need is an Upwork account—sign up or log in today to begin exploring how AI can transform your work.

‍

Disclosure: Upwork is a Jasper Affiliate and may receive referral payments from Jasper. When using Jasper, you will be subject to Jasper’s Terms of Service and Privacy Policy. As always, independent professionals remain responsible for evaluating the tools offered and determining the fit for their business needs, as well as their own compliance with all laws and legal requirements in operating their freelance business.

Upwork is an OpenAI partner, giving OpenAI customers and other businesses direct access to trusted expert independent professionals experienced in working with OpenAI technologies.

Upwork does not control, operate, or sponsor the other tools or services discussed in this article, which are only provided as potential options. Each reader and company should take the time to adequately analyze and determine the tools or services that would best fit their specific needs and situation.