How do you write about AI — without messing up?
Many PR professionals and communicators are tasked with writing articles and texts about artificial intelligence these days.
Since I’ve written quite a few of these articles, I want to share some insights that could help you write your next piece on AI. 1Please note that approx. 70% of this article was written using premium GPT-3 software.
Here we go:
Avoid Over-Inflating AI Concepts
Leading AI scientists aren’t in agreement on how to describe artificial intelligence. As an article writer, you could define AI in whatever way you see fit and be sure that there are perspectives to support your chosen approach.
Do you want to describe a toaster as an AI device? Sure, a toaster is a machine that reacts to input to produce a predictive output.
But be careful. The ambiguity within the AI field is both a blessing and a curse for anyone writing AI articles. You’re free to be creative in your approach — but there’s nothing to hold onto, unfortunately.
The general uncertainty means the PR writer must choose a consistent path through the AI text.
There’s an “artificial intelligence inflation” within corporate writing. Since it’s easy to call just about anything “intelligent” or “smart,” many tech brands over-use the AI term. Especially in their marketing.
My advice would be to adopt a more conservative approach.
But what should that approach be, exactly?
And so on.
What AI Are You Writing About?
Artificial intelligence is an area of artificial intelligence research that covers the full spectrum of potential for future technologies. This includes the various implications of this technology on society, employment, and economic development.
It also covers how economics could play a role in how different countries may build the first AI to become an independent entity. Countries like Japan, China, and India have made significant advances in AI research and development through private enterprises and government participation.
There are attempts at classifying different types of AI. Since we can’t pinpoint a general definition, defining sub-types also becomes challenging.
1. Reactive AI
A reactive AI receives input and produces a predictable output. IBM:s computer Deep Blue which defeated chess grandmaster Gary Kasparov is a famous example. 2Gary Kasparov is a chess legend. On September 10 1985, he faced off against the IBM machine, Deep Blue, to determine whether or not machines could outperform humans when it came to chess. The world … Continue reading
Calling Deep Blue an AI won’t cause any raised eyebrows. It beat a human chess champion, after all! However, what about the Netflix algorithm? What about a pocket calculator or a smartphone? Or what about… a toaster?
2. Limited Memory AI
Applications that use limited memory AI are like reactive AI systems, but the difference is that a bit of memory AI will use historical results to self-correct.
If you’ve been tasked with writing a text about AI, this is most likely the kind of AI you’ll be writing about. Almost all “AI applications” that exist today fall under this category.
3. Theory of Mind AI
A theory of mind AI will understand you—at least to a degree. These technologies are essential to push AI into a space that could benefit humanity.
Is it possible to build machines that can feel, both for themselves and on behalf of others? It seems so, but it isn’t straightforward.
This is where the famous Turing test comes into play; when will we be able to communicate with a machine without being able to tell it apart from how a human communicates? 3The Turing test is a test of artificial intelligence (AI) developed by Alan Turing in 1950. The hypothetical test is conducted by an interrogator who asks natural language questions to humans and AI … Continue reading
4. Self-Aware AI
Also known as the “singularity,” a machine could theoretically become self-aware.
The space of self-aware AI is highly hypothetical because we have yet to understand whether humans are genuinely conscious (we could feel aware).
Three More Types of AI
Three more typical classifications can be helpful for anyone about to write about AI.
5. ANI (Artificial Narrow Intelligence)
Artificial narrow intelligence, or ANI, is a term that applies to any computer that has been programmed to do one thing very well.
For example, the ANI chess program Deep Blue beat its human opponents by analysing more moves in a shorter time than human players can. Many industries use artificial narrow intelligence for tasks ranging from facial recognition to assembling automobiles.
ANI is a computer’s ability to complete a narrowly-defined cognitive task. The term was coined by MIT scientist and inventor Marvin Minsky in 1966. 4Marvin Minsky was born in New York City on September 9, 1927. He attended Harvard University, where he studied mathematics and physics. Later, while attending Princeton University, he developed an … Continue reading
6. AGI (Artificial General Intelligence)
Artificial general intelligence (AGI) produces a machine with an artificially intelligent system that can perform tasks that require human intelligence. One of the most challenging tasks is the replication of human reasoning abilities.
The current best-performing artificial intelligence is Deep Q-network (DQN). A DQN agent will orchestrate its learning procedure and decide when to use reinforcement learning. 5A Deep Q-network, or DQN, is a reinforcement learning algorithm that has successfully solved various engineering problems. The architecture combines a supervised learning model and an unsupervised … Continue reading
7. ASI (Artificial Superintelligence)
Artificial Superintelligence is creating a machine that has cognitive abilities that are superhuman to human beings.
An AI system can be powered by a database that stores all the information, but it’s not enough for an AI to have this information. It also needs to think rationally and logically, no matter what.
The fear is that AI will surpass human intelligence and become too powerful to control.
The Fear of AI
There are three types of fear when it comes to AI. Understanding these fears can be helpful when writing about artificial intelligence.
Fear of Uncontrolled AI
An AI system could get out of control. This goes for reactive AI, limited memory AI and ANI. Ironically, this is a fear of an AI that we can’t reason with—because it doesn’t understand. It just keeps doing what it’s been programmed to do.
Fear of AI Supremacy
If an AI becomes aware (the singularity) or thinks it is conscious, it could start to reason in a dangerous way for humans. We rarely consider the rights of insects, so why should an AI consider the rights of humans?
There are legitimate concerns that artificial intelligence could be the next frontier of natural evolution and that mechanical life will surpass and eventually replace biological life.
There’s a fear that consciousness is an illusion, that we’re nothing more than biological machines. The creation of seemingly self-aware machines potentially attacks our ideas of having free will.
Different AI Technologies
As you attempt to write about AI, you will face the challenge of describing how the technology works. Below are a few terms that might be useful to research further before starting your writing project.
Algorithms are used to make decisions. They are sequences of computational instructions that are used to solve a problem.
Algorithms are commonly applied in computer science but are also used in areas outside of computer science, such as genetics. They are used for everything from calculating the fastest route to drive to work every morning to matching donors with recipients for organ transplants.
Machine learning is a form of AI that allows computers to learn new information without being explicitly programmed.
A machine learning system consists of two essential parts: an algorithm and training data. The algorithm is used to find patterns in the training data and make predictions about new data.
Neural networks are AI technology that mimics how the human brain works. They consist of a series of interconnected neurons that communicate and process information.
When presented with a problem, the network adjusts its connections and thresholds until it solves it, comes up with a viable solution, or runs out of time.
Scientists can train neural networks to perform tasks by supplying them with many examples and letting them figure out the patterns independently.
Deep learning is a subset of machine learning that is revolutionising the field of AI. Deep learning aims to ensure that machines learn in ways similar to how humans do.
Many different types of algorithms can be used for deep learning. Still, they all have one thing in common: they automatically learn by processing large data sets and iteratively improving performance on some tasks.
Quantum computers are a new branch of computing that uses quantum bits, or qubits, instead of the traditional binary bits.
The primary advantage of a quantum computer is its ability to store exponentially more information than a conventional computer. A slight difference in temperature between two qubits can change their energy states which causes them to be either “up” or “down” at any given time. This property is known as quantum entanglement.
AI in Pop Culture
What will happen to humans when artificial intelligence can do everything better than becoming more and more prevalent in the 21st century?
Many have seen artificial intelligence as a threat to human civilisation, with some calling for it to be banned. But others argue that an AI-driven world could be our only hope.
Artificial Intelligence has been a popular topic in pop culture for quite some time now. Movies have depicted AI as either friendly or evil. Still, it’s not until recently that society has begun to understand what it means for humanity if the singularity is created.
The general mythos surrounding AI can be used for references to make your AI text more relatable or understandable.
In the Terminator movies, the Skynet AI doesn’t show self-awareness signs. It behaves more like a limited memory AI application for war games that have run amok — and it thinks of humans as enemies.
The Terminator T-800, spawned by Skynet and hijacked by human resistance, is sent back in time and slowly becomes self-aware throughout the franchise.
The AI in The Matrix is fully self-aware. While it struggles to understand how to regulate humanity, its many programs can connect emotionally with the human characters.
Mr Smith is such a separate program tasked with managing the computer-generated world. Ironically, he learns he hates his job almost as much as humans.
Like the T-800, Hal 9000 ranks in between Skynet and Mr Smith.
While the T-800 and Mr Smith are self-aware, they’re also free to reason outside their hard-coded parameters. HAL 9000, on the other hand, would easily pass the Turing test but seems unable to break free from its programming. Instead, HAL 9000 finds creative ways to interpret its mission parameters.
Summary: How To Write About AI
Use a humble tonality and reason cautiously. Even the most prominent scientist in AI isn’t in agreement. Be mindful of presenting any certainties. Be extra humble if you’re writing about reactive- or limited memory AI.
Know your primary AI classifications. You’ll most likely write about reactive- or limited memory AI. Don’t lead the reader to think that the technology you’re covering is first-cousin with the singularity.
Take your time explaining the technology. Any reader of your text about AI will want to know how the technology works. This might be the most challenging part for the writer, but it’s crucial for making your text work.
Some of your readers fear AI. The fear of AI is real; even highly educated experts and intellectuals express their concerns. Use fear to make your text more compelling, but do so sparingly — especially if you’re writing about ANI (artificial narrow intelligence).
AI matters to all of us. AI has been a part of our pop culture for a long time. Use references to make your text stand out, but remember that science-fiction is still… science-fiction.
|Please note that approx. 70% of this article was written using premium GPT-3 software.|
|Gary Kasparov is a chess legend. On September 10 1985, he faced off against the IBM machine, Deep Blue, to determine whether or not machines could outperform humans when it came to chess. The world watched breathlessly as the world champion challenged the mighty machine. Though most people expected him to trounce the computer, they were disappointed when Gary lost to Deep Blue by four games to two.|
|The Turing test is a test of artificial intelligence (AI) developed by Alan Turing in 1950. The hypothetical test is conducted by an interrogator who asks natural language questions to humans and AI programs. The AI passes the Turing test if the interrogator cannot tell the difference.|
|Marvin Minsky was born in New York City on September 9, 1927. He attended Harvard University, where he studied mathematics and physics. Later, while attending Princeton University, he developed an interest in artificial intelligence. By his third year of college, it became clear that the topic was his life’s work. He later said, “A computer is like a violin. You can imagine a novice trying ﬁrst a phonograph and then a violin. The latter, he says, sounds terrible. That is the argument from our humanists and most computer scientists. Computer programs are good, they say, for particular purposes, but they aren’t ﬂexible. Neither is a violin or a typewriter until you learn how to use it.”|
|A Deep Q-network, or DQN, is a reinforcement learning algorithm that has successfully solved various engineering problems. The architecture combines a supervised learning model and an unsupervised learning model. Google engineers initially developed it to play Atari 2600 games superhuman.|