The pioneers of computer science would be shocked by today’s AI.
Ada Lovelace wrote that a machine “has no pretensions to originate anything. It can only do whatever we know how to order it to perform.” To an extent, this is as true now as it was 200 years ago; Humans can create and perceive patterns from (apparent or real) randomness, because humans, for better or worse, see patterns everywhere. But computers can only work in existing patterns; it’s actually impossible for a computer to internally create truly random - and therefore truly original - data.
But the news is continually awash with stories of AI creating original art, poetry, and even paintings. So what’s going on here? Has AI advanced to the stage where it can be called truly “creative”? And how might we put it to practical use?
There’s no universal definition of creativity, just as there’s no universal definition of love, but we all know it when we see it.
So can a computer know creativity? More importantly, can we teach a computer to be creative? Recent advancements in AI seem to suggest that the answers are a resounding “yes.”
No human is born with abilities like writing, drawing, or singing—they’re skills we develop by observing and acting upon data, which we call “learning.”
Similarly, AI can observe and learn from data to gain creative skills like drawing, musical composition, and writing. Machine learning is the gold standard for AI, and involves quickly and efficiently training an AI model on far more data - whether that data is text, music, images, etc. - than a human could take in a lifetime.
The theory goes that if you feed an AI 100,000 poems, or 1 million pieces of pop music, the AI can analyze the patterns in them and then come up with its own. So today, AI has been used in a number of creative fields traditionally assumed to be “uniquely human”.
While famous artists from Vincent Van Gogh to Piet Mondrian died penniless, digital artwork has been sold for eye-watering sums.
To give a couple of recent examples, Sophia (a robot) sold a digital artwork for nearly $700,000, while the AI-generated Portrait of Edmond Belamy sold for over $430,000. More specifically, the portrait was developed using a Generative Adversarial Network, or GAN, which uses convolutional neural networks to mimic human creativity. Research shows that GANs can be made using RNNs, or recurrent neural networks, as well.
These GANs can become very accurate AI systems when fed with a lot of training data, as deep learning models tend to get better as they grow. Powered by huge clusters of Nvidia GPUs, GANs have been used to create “DeepFakes,” or realistic fake images and videos of people. These DeepFakes can be fed with new data in real-time, and since DeepFake models can be found online on places like GitHub, they’re not going away anytime soon.
Even if you don’t have a lot of training data, you can use machine learning techniques like style transfer to take advantage of pre-built AI models.
Even big brands like Nutella are using AI for design automation, at scale.
As reported in Futurism, Nutella used AI to create 7 million unique designs for its Nutella jars, all of which quickly sold out. Today, companies from Renault to Siemens use AI for product design, as reported by MIT Technology Review.
People have voted with their dollars: AI can draw, and it can draw well.
Not only that, but AI seems to be able to compose music.
As described by Google Arts & Culture, the song “Break Free” uses AI-composed notation and instrumentation, to viral acclaim, with over 2 million views. More precisely, this AI-composed notation used natural language processing, or machines that learn from text, rather than the computer vision-oriented models we looked at earlier.
Back in 2016, IBM used AI to make the movie trailer for Morgan, a budget film that went on to a nearly $10 million box office.
Writing poetry is notoriously difficult, even among creatives. As Paul Engle wrote in a New York Times article, “poetry is ordinary language raised to the Nth power.”
At the same time, AI-generated poetry is now able to fool humans, as reported by the World Economic Forum.
Evidently, AI can appear to be creative, if you view creativity as just another skill that can be learned from data.
Given the tremendous rate of progress we’re seeing in AI, we can naturally expect AI’s creativity to continuously improve to the point where AI-created art could be indistinguishable from human art in any area.
But is this a satisfactory answer? The concept of “artistic” creativity is deeply philosophically contentious, and hinges on twin issues of randomness and understanding.
As we’ve discussed, computers are currently unable to create truly random data. And some people might argue that it’s that element of randomness that is itself the spark of human creativity - adding something that no-one else could or has added before. An AI cannot do that.
Similarly, although AI may be able to fool a human into thinking a given poem, painting, etc., was written by another human, that’s not the same as that piece of “creative art” having genuine meaning. An AI will give you an endless number of outputs based on what you put in - but it won’t understand them. It has no concept of why you should be compared to a summer’s day, other than that string of words appeared in another piece of text it analyzed. To an AI, all creative inputs and outputs are merely data.
But that’s not necessarily a bad thing, because innately-creative humans can work with that data.
AI can provide a platform for humans to augment and strengthen their own creativity.
For example, AI is great at doing lots of things, quickly. It can come up with many ideas, concepts, or rough drafts, which a creative person can use to expedite and inspire their own creative process.
Sure, AI alone may not create the end result, but it can act as a co-creator.
In the examples previously outlined, humans had at least some role, even if it’s just selecting the subjectively best artwork created by an AI.
Since AI doesn’t know what humans enjoy, it’s still up to creative humans to figure out what will resonate with other people. In short, humans still need to discard the uninspiring creations and use their own creative sense to improve upon the best creations.
Akkio can augment creativity in a number of ways, including for SEO and copywriting, forecasting, and the creative arts.
While we’re still a long way off from AI being able to convincingly write article-length prose, AI can still help in a number of ways.
For example, Akkio can be used to tag content with the most relevant keywords, such as "advertising," "branding," or "generating leads,” using text classification.
In the past, marketers relied on manual work to tag articles with keywords. This process was time-consuming and distracted marketers away from more creative work.
Now, marketers can use AI and automate this process.
As Neil Patel writes, there are now a number of AI tools that even assist with the writing process itself, such as CopyAI, Wordtune, and Copysmith.
These tools help marketers overcome the dreaded “writer’s block,” and therefore boost efficiency. Even if the AI-generated copy isn’t used directly, it can help propel the creative process. Remember, although AI can create reams and reams of prose, it doesn’t know what it’s talking about - which is why skillful and knowledgeable writers are still needed as a core part of the writing process.
In other words, AI can be a powerful co-writer. After all, writers have been using less-technical solutions like auto-complete for years, and today’s AI copywriting tools are just the next step in the evolution of a writer’s arsenal.
Moreover, AI can identify and score leads, ranking them based on their likelihood to convert.
Leads are scored based on past purchase behavior or decisions they have made during the buying process, such as providing an email address or phone number.
The lead score may also be based on more subjective elements, such as the individual's social media presence, web browsing history, or physical location.
A higher lead score indicates a higher probability of conversion and, in turn, a higher potential ROI for marketers. Conversely, a low lead score means a lower probability of conversion and a lower chance for ROI.
The use of scoring means that marketers are able to allocate their time towards targeting the most promising prospects with tailored content. By automating the logical, numbers-based process of lead scoring, marketers have more time for creative, personalized efforts.
Good customer support is critical for turning customers into promoters, minimizing detractors, and reducing churn.
As Hubspot writes, creative problem solving is key to stellar customer support, as many support tickets will require agents to be creative and come up with unique solutions. Hubspot re-tells the famous story of a Ritz-Carlton guest accidentally leaving behind their laptop charger. Without hesitating, Ritz-Carlton sent a next-day air package with the charger to the astonished guest.
For most businesses, this level of customer support just isn’t possible for every single customer. Most customer support teams are already stretched thin, so they can’t go above and beyond at every possible turn.
AI-powered ticket scoring solves this problem. By using AI to rank customer support tickets based on their importance, support teams can prioritize their work.
This informs customer support teams what tickets are worth focusing on, so they can come up with creative, personalized solutions for the customers that make a difference.
Classifying customer support tickets based on their importance uses a type of AI called text classification. The same technique can be applied to any type of text, including social media posts.
Nowadays, millions of consumers turn to social media to voice both their criticism and commendation of products and services. Businesses that quickly address concerns, and show gratitude for happy customers, have a huge edge over their competitors.
As with support ticket classification, this will give teams the time to provide creative solutions for customers in need.
Moreover, text classification on social media can be used to find dissatisfied customers who are using your competitors’ products.
By building an AI model to, for example, classify the sentiment of tweets, you can find out when users are disappointed in a competitor, and deploy creative outreach messaging.
Marketing is about so much more than just lead generation, as conversion rates are just as important. To go a step further, marketing teams need to analyze existing customers and prevent churn before it becomes a threat to the business.
In fact, churn is one of the leading killers of businesses, as any churn rate that’s higher than the growth rate will inevitably lead to business failure.
There are so many factors that may impact churn, and the probability varies from one customer to the next, which makes manual churn prediction infeasible at scale.
With AI, businesses can automate the process of churn prediction, so they can flag customers likely to leave next, and find creative solutions to keep those customers on board.
As the world becomes more integrated and interconnected, we are seeing a major shift in how we innovate.
AI forecasting is one of the new tools that is helping us plan ahead and come up with creative ideas.
AI can synthesize data from disparate sources to come up with a forecast of what might happen in the future, and then suggest actions that could be taken to maximize success or ward off potential problems.
The forecasts are generated by algorithms that scan vast amounts of data, whether it’s from social media, mobile phone usage, weather patterns, or even changes in the stock market, looking for patterns and correlations.
With Akkio, forecasting is effortless, giving businesses the awareness and time to creatively plan for the future.
Just because an AI by itself is unlikely to come up with artistic gold, it doesn’t mean that you can’t work with it to power up your creative imagination.
As we’ve explored, artists have fed poetry, music, art, and more into AI models, and created their own art with the result. The deeply creative parts are for you to think of!
Crucially, with Akkio, AI is open to anyone. AI is no longer the preserve of big tech companies. Anyone can build and train a model on Akkio in minutes, with a free trial and low prices with unlimited usage beyond that.
Akkio exists to encourage experimentation and new uses for AI, democratized and available to all. Let your creativity run wild and see what AI can do for you.
Artificial intelligence is often thought of as a purely logical, mathematical field. Today, however, AI is being used for creative applications in fields from marketing to music.
For example, marketers who want to better understand their potential customers can use machine learning to create customer personas based on datasets. Akkio’s Salesforce integration allows marketers to easily build customer personas or segment customers by demographics.
Similarly, developing a new piece of music might entail a computer studying pitch and frequency of different musical notes. AI can even be used to draw or write poetry. With breakthroughs like DeepMind’s AlphaGo and OpenAI’s GPT-3, there’s surely a bright future for creative AI.
AI’s greatest potential in this area is not replacing creative humans, but augmenting their work. By automating repetitive and tedious tasks, AI can even free up time for businesses to come up with more creative solutions, whether that’s to improve customer satisfaction, improve lead conversion, or otherwise improve the bottom-line.