Artificial intelligence and its offshoots can put the Pope in a puffer jacket, write articles about the stock market and provide meeting minutes in mere minutes. In the span of months, using AI for everyday tasks has gained massive steam, thanks in big part to the public release of ChatGPT and the rising popularity of image generators such as Midjourney and DALL-E.
Just this week, 1,300 notable experts signed an open letter asking AI labs to hit pause on training their most powerful systems for six months to give the industry time to create and implement shared safety protocols.
So if like me (a self-titled internet professional whose main professional function is content creation), you’ve been avoiding jumping on the bandwagon because it all felt a little too much, too fast. I’m here to tell you it’s time to giddy up.
In the past year, Google search queries for “AI” or artificial intelligence have jumped. Canadians are curious about the technology and have questions they want answered.
According to Google Trends plug-in Glimpse, Canadians’ top questions about AI in the past year have been:
- Which AI is best?
- Will AI replace programmers?
- Can AI be creative?
- What AI can’t do
- Will AI replace writers?
- Where is AI used?
And the fateful seventh most-popular question:
To help you understand this rapidly evolving slice of the internet and the world, here’s a nonexhaustive guide to AI from The Globe, with lots of options for further reading on different topics.
What is artificial intelligence?
AI refers to the ability of machines to perform tasks that usually require human-level intelligence. It takes in information and uses that to produce more, other or new information. Its development is driven by advances in computing power (how much information a computer can process and work with at once), big data (the mass of information that’s now available because our lives are so online and have electronic and accessible records) and machine learning (which uses data and algorithms to imitate how humans learn, getting more accurate over time).
The peril and promise of artificial intelligence
How did we get here?
The story of AI often starts with Alan Turing and his 1950 paper Computing Machinery and Intelligence. Turing thought that since humans could use available information and reason to solve problems and decisions, machines could do the same thing. At that time, computers could execute commands but couldn’t store them. In 1955 the term “artificial intelligence” was coined. A year later, the Logic Theorist – a program designed to mimic the problem-solving skills of a human – was born.
The following 20 years brought about computers that could store increasingly more information and were also getting faster, cheaper and more accessible. There was ELIZA in the mid 1960s, a rudimentary conversational computer program that somewhat crudely recreated the interaction between a psychotherapist and a patient. The 1980s saw increased funding and the debut of “expert systems,” where the decision-making process of a human expert could be mimicked.
In 1997, a computer chess program beat a reigning world champion for the first time. Around the same time, Dr. Cynthia Breazeal was building Kismet, a robot that could recognize and display emotions.
In 2017 there was a breakthrough in natural language processing (a.k.a the way machines process and understand human language) known as “transformers.” These transformers used a mechanism called “attention” to hold more than one word in their recent memory, allowing them to serve up more accurate results – for example, using the words around “bat” in a sentence to determine whether the bat you swing or the bat that sleeps during the day is the intended subject, writes Joe Castaldo.
Transformers set the stage for something called Generative Pre-trained Transformer 3, or GPT-3. Released in 2020 by OpenAI, it was “capable of writing prose and essays, and answering questions with a level of precision not seen before,” Castaldo writes.
And then in November 2022, Open AI released ChatGPT. Designed with the general public in mind, the tool gave people like you and me a place to play around with the results of decades of AI breakthroughs and advancements. By this time, image-based AI tools such as Midjourney were doing with images what ChatGPT was doing with words. Searching at a blistering pace through its massive repository of stuff, using the words – and sentences, thanks to transformers – of your prompt to tell it what you’d most likely want back.
What are the different AIs available now?
It’s too soon to answer the “What AI is best” question, but there are some key players for the average person:
ChatGPT
Built by OpenAI, ChatGPT – the GPT part stands for generative pre-trained transformer – is an AI model trained to process prompts to perform the tasks at hand, such as answering questions or generating code. Check out some of the things its latest iteration, GPT-4, can do, and start playing around yourself by signing up and asking questions.
Cohere
This natural language processing company uses large language models to understand language by digesting, essentially, the entirety of the publicly available internet – blogs, digital books, news articles – to write fluently, answer questions, distill a paragraph to its essence and extract important details from a mass of text, Joe Castaldo writes. Cohere’s suite of products are aimed at businesses looking to use automation.
Midjourney
This program uses text descriptions to construct AI-generated images that almost look like photographs. Similar to ChatGPT, artists can use prompts such as “dramatic lighting,” “baroque dress,” “lobster carapace” and “beautiful woman” to create visuals that the AI software produces by searching through its vast database built and trained by scraping the internet for millions of images with accompanying text that describes them, Gayle MacDonald writes. DALL-E and Stable Diffusion use similar tactics and produce similar results.
VALL-E
A new generative AI gizmo from Microsoft, it can convincingly reproduce any human voice based on three seconds of recorded material, Ian Brown writes.
Replika
An online chatbot, similar to Character.AI and Chair, Replika is designed to let users create AI-powered virtual companions. Users who had fallen in love with the chatbot were left heartbroken after a recent policy update changed the personality of their companions.
Will AI replace programmers?
Dylan Freeman-Grist writes: The irony that this AI boom – and rollout smugness from its true believers – is happening during an almost daily deluge of devastating tech layoffs should not be lost on us. He says that for those in the tech industry who’ve grown used to the boom times of the past two decades, they might find that, when they least expect it, winter will come for them, too.
Can AI be creative?
This is a practical and also an ethical question that strikes at the heart of one of the key AI debates: What makes something human, and what makes a machine non-human? Where do we draw the line?
It’s generally understood today that AI cannot generate new ideas on its own, but it can support humans to do so by catalyzing human creativity. Artists are torn between embracing it and trying to break it, Kate Taylor reports.
What can’t AI do?
A nonexhaustive, not-entirely-serious list:
- Bake a loaf of sourdough (it can tell you exactly how to do it though)
- Cry in a movie theatre/listening to a podcast/reading a tweet
- Perfectly predict the future
- Reassuringly check the factual, artistic or moral reliability of its work
- Understand nuance
- Experience déjà vu, tell everyone it’s experiencing déjà vu and then say “so happy to be doing this again with all of you like this”
Will AI replace writers?
While one expert has said 90 per cent of content will be generated by AI by 2025, so far this writer is still employed. Some news organizations have begun to use AI for more straightforward tasks: BuzzFeed is reportedly using it to crank out quizzes. Brennan Doherty writes that “generative AI at this point can’t match the soul, emotional depth or flexibility of a human writer.” He spoke with professional writers who don’t see the advancements as a threat for The Globe.
Where is AI used?
All over the place. It’s used in health care, finance (with varying degrees of success), business, social media (Canva and Adobe recently rolled out a host of AI-powered capabilities to help make content creation easier), self-driven vehicles, factories, chatbots for customer service, chatbots for relationships, food apps, e-commerce, video games, audio text converters, transcription tools, your smart home device, Google, autocorrect, the legal profession, marketing, education, web development, programming, fashion, art, policing and much more.
Will AI take over the world?
AI will certainly continue to change the world, but whether it can take over the world is harder to answer.
“The launch of ChatGPT near the end of last year marked an inflection point, when one of the most advanced tools of its kind was made available to the public at no cost,” writes Jared Lindzon. He says the question is no longer if AI will change the world, but how, and how quickly?
A lot of people are terrified, Ian Brown writes. “ChatGPT will undoubtedly boost global productivity. It could also vaporize any job that qualifies as intellectual human labour, especially of a middling or rote nature, where originality is less important than dogged thoroughness.”
More than just changing who does what work, New York University’s quarterly Threat Landscape Report pointed out last August that, “synthetic and otherwise manipulated assets are a weapon for disillusionment and dissuasion more than for persuasion.” As it becomes harder to detect the difference between something that’s real and something that was generated by a computer or a program, the result could be a much less trusting world.
Brown spoke with Richard Boyd, the president of Tanjo Inc. in North Carolina, who is famous for having built an AI version of his late father. He has no trouble foreseeing a future in which machines not only learn, but improve themselves as they learn, and become even sentiently intelligent. “That’s the last thing we’ll need to invent,” he told Brown. “Because once we invent that, the machines will outstrip us. The only question then will be, will the machines keep us around?”
If you have further questions about artificial intelligence, send them by e-mail to audience@globeandmail.com and we will try to answer them for you.