Generative AI is a technology that is ready for takeoff in 2023.
It’s been years since we’ve seen a technology that’s so captivating, with the ability to disrupt work as we know it. Twitter is inundated with screenshots of people prompting OpenAI’s ChatGPT to write wacky stories about Wayne Gretzky playing water polo or MidJourney creating a photorealistic carbon copy of someone’s beloved Golden Retriever.
Venture capitalists (VCs) are frothing at the mouth in a downturn economy at the prospects of generative AI. According to PitchBook, VC investment has skyrocketed by 425% since 2020 to $2.1B.
But what exactly is this new technology, and how can you use it to help you in your business or job? And how will it impact the world over the next decade and more?
To find out, keep reading below!
Generative AI refers to artificial intelligence that can generate unique content. Using machine learning, generative AI models can create basic software code, art, prose, text generation, and much more.
Generative AI has exploded onto the scene recently with the release of OpenAI’s ChatGPT. This has opened up a world of possibilities for those who want to utilize generative AI capabilities.
In this article, we’ll look at some of the pros and cons of using generative AI, some examples of generative AI in practice, how generative AI works, and what the future of generative AI may look like.
Let’s dive in!
Generative AI typically uses unsupervised learning or semi-supervised learning to process vast swaths of data to generate original outputs. For example, if you wanted your AI video generator to create paintings like Andy Warhol, you’d need to feed it as many existing Andy Warhol paintings as possible. Similarly, if you wanted your generative AI tool to write like JK Rowling, you’d want to feed it as many Harry Potter books as possible.
The neural network that generative AI models are based on will take all those images or prose and generate new material based on what it has learned. Generative AI systems are designed to be able to create new content that is both unique and useful in its own right.
Generative AI typically uses a particular type of transformer called a GAN to generate content.
According to Google, GANs (generative adversarial networks) are a special machine learning model that can create new data instances resembling a model’s training data.
GANs pair a generator (which learns to produce a target output) with a discriminator (which learns to distinguish true data from the generator’s output). It’s like a game of cat and mouse – the generator is always trying to fool the discriminator while the discriminator tries to keep itself from being fooled.
The discriminator continually attempts to distinguish real-world data from the fake data generated by the GAN’s generator model. Every time the generator succeeds in “fooling” the discriminator, it gets rewarded, hence the “adversarial” model. This allows the generative AI model to train itself repeatedly on new training data, generating more realistic content as more data is fed into it.
If that sounds confusing (it still is confusing for me), here’s a video from IBM to help you learn more about the basics of how GANs work.
Let’s look at the three main types of AI generative models below.
The rise of the aforementioned ChatGPT has brought generative text AI to the masses. Text-generative AI tools can generate text-based content based on a user’s queries or inputs. This means that you can type a query into a text box, and the machine learning model will analyze the text and generate a response based on what you typed into that text box.
This includes, but is not limited to:
- Blog Posts
- Software Code Generation
- Creative Writing
Basically, if it’s something you can write, the generative AI outputs can write it too. This is quite handy for scenarios where you need to generate a lot of content quickly, whether social media content or blog posts. Simply tell the generative artificial intelligence model what you’d like it to write, and it will spit it out for you.
This helps ensure you don’t get stuck if you ever have writer’s block. If you’ve ever found yourself unable to finish a sentence or thought, text generation can finish it for you, allowing you to complete the thought and continue writing. This is instrumental in ensuring that you continue to write and get words down on the page, allowing you to finish your project faster and more efficiently than ever before.
They’re also great tools to utilize in your creative process if you need help developing ideas for content, headlines, and anything else you can think of. For example, I used Jasper’s Chat Function to generate 30 different heading ideas for this blog post, and I picked and iterated on the one I liked most.
Generative AI art is disrupting the traditional art market with its ability to generate unique, stunning visuals. Generative AI art tools use deep learning techniques to produce photorealistic photographs or other artwork.
Simply enter what you’d like the generative AI system to draw, and it will spit out a shockingly high-quality image in various styles that you can choose from.
You can even ask the software to create a specific style, for example, something more photorealistic or abstract, and it will create an image as you requested.
Some of the uses of AI art include:
- Creating videos for social media
- Generating logos for businesses
- Designing book covers and album art
- Producing visuals for websites or other marketing materials
- Making graphics for video games or virtual reality
Although the technology is not as far along as text or image generation, video generative AI systems are emerging and becoming increasingly popular. With more time & development, AI video generators will become increasingly more powerful, perhaps one day disrupting the entire video medium.
These systems can generate videos from text or images, allowing users to create unique videos quickly and effortlessly. It’s perfect for those who want to create high-quality video content without having to invest in hiring a professional video production company.
Generative video AI has yet to be powerful enough to replace traditional video. Still, it’s not difficult to see where the technology is headed, especially given how quickly AI art and text generators are advancing.
Some of the uses of video AI include:
- Creating explainer videos
- Generating video content for social media
- Producing visuals for promotional materials
- Making short films or documentaries
As you can see, the above use cases cut out many key parts and roles in the video creation process, including video editing, scriptwriting, voiceovers, acting, and more.
Generative AI has a lot of advantages over traditional methods of content creation. Here are some key benefits:
- Speed: Generative AI can generate massive amounts of content at the click of a button. While it is lower-quality content than a professional could create, it is far quicker to generate text utilizing large language models than having a human manually type something out.
- Automation: For busy entrepreneurs, generative AI models have been a godsend as automating more content generation allows them to focus on more pressing or strategic aspects of their business.
- Cheaper: Business owners utilizing generative AI effectively can save money by not hiring human writers. A typical AI text generator subscription costs around $50/month. The tradeoff is lower-quality prose, but depending on your content strategy, that may be a sacrifice you’re willing to make.
- Unique: Generative AI can generate unique pieces of content every time, making it great for producing one-off pieces. The ability to create unique content is helpful because it gives writers or artists the power to share with their audience something they’ve never seen before without the hassle of manually creating something.
Some of the disadvantages of using generative AI include the following:
- Lower Quality: Simply put, generative AI models are not at the point of surpassing human writing. The quality just isn’t there, hence the common complaint that AI-generated text sounds like it’s written by a robot (not shocking, because it is). Another example is AI video; the technology is nowhere near replacing cartoons or B-footage.
- Limited Creativity: Generative AI creativity is limited by the data fed into the model. With less training data, some machine learning AI models spit out low-quality text that doesn’t make much sense in the context of a more extensive article.
- Lack of Human Touch: Generative AI models do not possess humans’ higher-order cognitive abilities. While they can automate specific tasks, they still need to gain the creativity and thinking humans can bring to the table.
- Inaccuracies: Generative AI models are fraught with inaccuracies. It only sometimes gets that right if you ask an AI art generator to draw a cat with four legs. If you ask ChatGPT a question, it often spits out the wrong answer. This is because, as mentioned before, AI generative models are based on training data. Sometimes, the AI models aren’t robust enough for the generative AI to create high-quality or accurate content.
- Eliminating Human Positions: AI art is becoming so realistic that it’s removing human artists’ ability to make a living. Why commission an artist to draw an immaculate painting of you when you can ask Midjourney to generate eight photorealistic photos for only $30/month.
With new technology, there are always risks to be aware of – especially with how powerful AI technology could one day become. I’m not *specifically* referring to robots taking over the world here; there are real-world consequences for improperly utilizing generative AI, specifically, that everyone needs to be aware of.
Some of the risks of generative AI include the following:
- Deepfakes: Deepfakes are AI-generated videos and images that can manipulate the public’s perception of a person or company. They are shockingly realistic looking. Imagine a deepfake of a powerful politician spouting off disinformation, influencing large swaths of the population to take action based on what they’re saying, leading to disastrous real-world consequences.
- Disinformation/Misinformation Campaigns: Unsurprisingly, many bad actors worldwide are chomping at the bit to utilize generative AI to create misinformation campaigns. This could be anything from fake news articles to automated comments on social media videos spouting off propaganda that isn’t actually true.
- Elimination Of Human Jobs: As mentioned earlier, generative AI can replace human jobs, leading to many people being out of work. That’s never a good thing for someone that has mouths to feed.
- Proprietary Content: Because AI-generated content is based on existing, human-generated content, as the generative AI industry expands, there will no doubt be ethical questions as to what content is truly “original,” what is “proprietary content” and perhaps a re-evaluation of how content ownership is defined today.
As discussed in this article, AI has come a long way in the past few years. Generative AI applications are transforming how content creators approach their work, whether in writing, art, video, and more.
The space is already becoming extremely saturated with many different text, video, and art generators already on the market.
Such technology buzz is bound to attract venture capitalists. One of the most popular AI text generators today, Jasper.AI, recently secured a $125M Series A funding round at a $1.5B valuation, anointing its unicorn status in a down economy.
Similarly, in January 2023, just a couple of weeks ago, Microsoft and OpenAI extended their partnership with a multi-year, multi-billion dollar agreement to continue to build and develop safe AI systems.
In fact, Acumen Research and Consulting predicts the generative AI market size to balloon to $100.8B by 2030.
Lastly – there are many other exciting applications for generative AI in many industries that we haven’t discussed. According to IBM, “66% of all companies are either currently executing or are planning to apply AI to address their business goals.”
Well, there you have it. We’ve covered a lot of ground here, including a definition of generative AI, its pros and cons, risks and benefits, some current use cases, and how it could be used in the future.
One thing’s certain – if you’re not keeping up with generative AI, you’re missing out on a once-in-a-lifetime technology. It’s one of the most compelling new technologies in years, with the ability to supercharge your business & work if used correctly.
Generative AI has breathed new life into content creation, and I can’t wait to see its continued development and utilization in the coming decade.
Have any use cases I missed or any comments you’d like to add? Feel free to comment below; I’m always happy to chat!