Last year proved a formative year for technology. Businesses are becoming increasingly exposed to different forms and applications of automation – from machine learning to natural language processing (NLP) – while also now having to consider the ethical and legal implications. Conversations are happening across the enterprise, with systems now competent enough to automate some core back-office functions.
One of the most exciting developments, however, has come in the form of generative artificial intelligence (AI), made famous by the likes of ChatGPT and DALL·E 2. This is a breed of AI that creates, combines and remixes content in the name of creativity. Generative AI can produce imagery, and written words as well as even mimic the human voice, sparking widespread discussions around the ethical implications and whether there’s any scope for AI replacing core creative functions in people. Much of this depends on the specific tools out there, though, as well as the quality of the product and whether people will be minded to pay attention to the potential drawbacks.
What are the origins of generative AI?
Although generative AI was made popular last year with the expanded use of the likes of MidJourney, conversations around this technology have been happening for much longer. Dutch researchers wrote about the philosophical underpinnings of generative AI as far back as 2012. Indium Software, too, released a white paper less than three years ago which highlighted not just how generative AI could be used creatively, but also in high-friction workplaces like healthcare.
For many it’s the natural evolution of creativity and tech. These functions are inclined to morph into engines that can garner the type of appeal we’re seeing now, the thinking goes. This is true not just for those heavily steeped in the IT or enterprise world, but the average social media user who’s curious about this previously opaque process. With the momentum this technology has, it’s no surprise recent Grand View Research findings put the generative AI market at close to $109 billion (approximately £91.4 billion) by 2030.
How does Generative AI work?
Generative AI is a process that can take a few forms, but at its core, it means using tools like neural networks – in the form of Generative Adversarial Networks (GANs), transformers, variational auto-encoders, and diffusion models to take inputted creative data and spit out new creations. We’re using the word creative broadly here to not just mean your friend who’s trying to up their graphic design game with minimal resources, but also the innovative uses available in business settings.
As opposed to the form of AI that attempts to categorise information, generative AI relies on modelling that tries to understand the dataset structure and generate examples that might relate, or match. The two main forms of neural networks at play here – GANs and transformers – work in slightly different ways. By and large the former is involved in creating visual and multimedia content from images and text, while the latter uses information on the internet to generate textual output.
What are some examples of generative AI?
There is a litany of tools available now that are powered by neural networks, ranging from image-based tools like Midjourney and DALL·E 2 to text-based systems such as GPT-3 and LaMDA, as well as the forthcoming GPT-4. There’s also VALL-E, developed by Microsoft, which is a tool that’s used to generate voice.
Beyond attempts to make Christmas cards, tools like Open AI’s DALL·E 2 and ChatGPT have been used to explore their impact on other creative work. There’s the added influx of AI-generated profile pictures on social media, using Lensa AI, while tools like GPT-3 have been experimented with by YouTubers to generate video content for the streaming service. Corridor Digital, for example, repeatedly explored how generative AI may disrupt video production by mimicking special effects.
What are the benefits of generative AI?
Keys to successful innovation through artificial intelligence
Harvard Business Review
Part of the reason that generative AI has seen such a boom in popularity is its eye-catching output, ease of use and capacity for wonder. Perhaps someone is looking for the AI-assisted version of a collage to spark their creative process, or maybe they’re curious about what suggestions an AI model has for a resignation letter. For those seeking a new challenge in 2023, they might lean on the machine to better advertise their skills and capabilities. Some people might simply wonder what would happen if their two favourite artists collaborated.
On a more practical – and much less ethically dubious – level, Indium Software suggests generative AI might be used in the healthcare sector to make images more usable. Automation company Fireflies.ai, meanwhile, suggests generative AI can be used effectively for things like large-scale logistical planning, and identity verification at airports. Still, each of these use cases creates more questions than it answers. There’s no doubt that powering these tools, and using them, will involve a tonne of data – so where will that data be stored? There are also privacy issues and cyber security concerns around sensitive use cases.
What are the main concerns around generative AI?
One risk of generative AI that’s already been widely discussed is the potential to exacerbate the rise of deepfakes. These images or videos are created in such a way that renders a lifelike imitation of a person, often a celebrity but also potentially a prominent business leader, in such a way that it can trick others.
There’s widespread concern that technological advances can make this threat really sophisticated, especially in the way that generative AI can be weaponised. For example, real-time deepfakes are now becoming a more serious threat. Tools like Microsoft’s VALL-E – an AI-powered system that can mimic the human voice with only three seconds of input audio – might spark a new era of cyber crime.
Reports have emerged recently of how using AI assistant tools in code can post significant security risks. Generative AI, with its heavy focus on creativity, can pose similar problems. There are many examples of someone feeding a tool like MidJourney a prompt and seeing a distorted copyright logo in the corner. These tools are not particularly discerning and MidJourney CEO David Holz admitted to Fobes that permissions were not sought from artists for the works that their product was trained on.
Another concern is related to education. Teachers at all levels – from early years learning to industry certifications – are anxious that ChatGPT and future iterations like GPT-4 can affect how qualifications are achieved. If you Google “AI Plagiarism Tracker” you’re likely to get just as many results as you might expect for the tools themselves. There are those who suggest cheating has always been possible, and that the number and quality of tools at hand aren’t relevant, but the concern is still notable with the picture continuing to evolve.
While it’s tempting to see generative AI as a maligned force, given the early chaos it has sewn in the creative industries and across parts of the economy, that doesn’t mean it can’t be tamed and channelled into productive use cases. That said, there are many questions that need to be addressed first, with regulation a hot topic at the moment. The AI Bill of Rights, for example, was drafted in the US as a framework for shaping future regulation. There’s no doubt that any future AI regulation will aim to devise a set of rules and regulations around the use of generative AI to ensure it leads to the maximum level of benefit with the least amount of harm possible.
Get the ITPro. daily newsletter
Receive our latest news, industry updates, featured resources and more. Sign up today to receive our FREE report on AI cyber crime & security - newly updated for 2023.
John Loeppky is a British-Canadian disabled freelance writer based in Regina, Saskatchewan. His work has appeared for the CBC, FiveThirtyEight, Defector, and a multitude of others. John most often writes about disability, sport, media, technology, and art. His goal in life is to have an entertaining obituary to read.