Nowadays Artificial intelligence (AI) has become a hot topic and a widespread buzzword, by more and more companies that are trying to grasp benefits of what this technology has to offer. But what exactly AI is and how does it work? Is this technology a new one or just an updated version from history? How to use it and how in particular it could help such industries as digital marketing, art, music and alike?
What Is AI and How Does It Work?
To begin with, the technical leap in AI came after the “revival” of old computational methods in 2012, with the publication of a paper on neural networks and their potential by University of Toronto researchers Krizhevsky A., Sutskever I., Hinton G. E. This single yet very important paper was like a fresh air to the word, but as usual, various names of technology, such as Machine learning and Deep learning, has spread in mass media, making some confusions. So firstly, let’s take a brief look on what these keywords mean and what the main difference is between them.
Artificial intelligence is a name of the whole technological sector where computers are programmed to learn, solve problems, and mimic the actions of humans. Machine learning is a field of AI that addresses a wide range of practical issues without providing a definitive computational algorithm. In machine learning, the algorithm learns itself from the initial data provided, meaning that it provides possible solutions, and when the data is updated, the computational model is updated also, and improved solutions are provided.
Artificial neural networks and Deep learning are just subcategories of AI methods and techniques which serves better for particular problems we want to solve or products to generate.
In general, the application of such learning systems (AI and Machine Learning) offers advantages beyond the old mathematical models of computation where success of the process is strictly defined of how mathematical formulas and predefined logic performs for a particular task. If your code is somewhat incomplete, you will get bad results. AI techniques, like Deep learning, changes that enormously by letting to perform tasks intelligently without being explicitly instructed to do so. And that, ladies and gentlemen, is a great power!
Nowadays AI allows to manage the computational processes, to see how the computer chooses solutions, and to update the initial database when necessary. The most important word here is database. This means that to teach computer to classify images, create new computer-generated picture, solve puzzle, or propose possible outcome of any kind of a problem, firstly we must collect a big database with examples, elements and already known solutions.
For example, if the goal of AI is to recognize a car in any photograph, it is first necessary to analyse in detail what the car looks like when photographed at different angles, in good and bad lighting conditions, etc. The system needs to be trained to recognize cars not from the capture/non-capture of the wheels in the photograph, and to decouple lighting and colour from the assessment of the silhouette, etc. This will allow the recognition of a car captured under different conditions and increase the recognition rates, meaning AI will be powerful enough to identify a car in any photo even if the car is partly hided or only some part of the vehicle is understandable for the human eye.
This example illustrates one of the biggest AI superpowers – ability to identify objects when sometimes it is impossible to do so for a well-trained human. What is more, AI can do it with a better accuracy and much faster. Simply, AI helps various organizations and individuals to make better decisions, enhance products and business’ processes to become much faster.
There are a lot of different kinds of AI categories, stages, or theoretical stand points, like Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI), Artificial Super Intelligence (ASI), Cognitive Computing AI, Computer Vision AI, Strong/Weak Artificial Intelligence, to name a few. Not to mention learning methodologies like supervised learning, unsupervised learning, reinforced learning, semi-supervised learning, and so on.
These concepts go to a very deep rabbit hole which is not a topic of this article, rather is just a demonstration that AI consists of various possible elements and methods, but general idea is quite clear – we must build a big database, train computer models, test which parameters and methods work best for our case and constantly improve the process.
AI in Creative Industries
While there are tons of applications where AI helps to forecast weather, make search engines better or diagnosing diseases, some part of AI improves industries such as product design, music, copywriting, and digital marketing. How can AI be beneficial here? What are certain tools already bringing new possibilities?
In the near future AI will make a much easier communication between computer power and brain power. Today when we need to make a sketch of the new magazine’s cover or to model 3D object for a game scene, we press a bunch of options, keys, and sliders in order to get a result what is understandable for us but not for a computer. AI will improve a connection by understanding what your final goal is and what you really want to see.
Especially sketching can be fun and productive in accordance with some VR (virtual reality) tools like Gravitysketch. It does not use AI at the beginning, however we have to admit that rapid modelling in VR with virtual reality glasses, in real time and using some AI creatively generated ideas could easily bring some valuable results to the work pipeline. For example, to sketch a prototype of a shoe in VR and an hour later print it with 3D printer could become a “new normal” in creative industries such as product design.
VIZCOM lets you test your hand sketches into the rendered stuff which will be more realistic and understandable to go further in the design workflow. That could, of course, save valuable time, but what if a designer has lost his/her ability to produce any news ideas to begin with? Possible answer could be just “have a break”, “get some rest” and “get back when you are ready to be creative”.
Or…you can try to use such AI tools as Artbreeder. This tool lets you create hundreds of variants of images of any kind. Let’s say you have to illustrate some article with non-existing human. Just generate it with Artbreeder and set preferred parameter of age, gender and so on. Maybe you need a game character or some non-existing landscapes? That tool could generate interesting ideas ready to be used as it is.
Nutella used some AI engines to create 7 million different jar designs and it was a successful project. To design all such images would have to take many workers and many days if not months to produce random design. Hence such AI helpers could be a real game changer for projects like product personalization and quantitative challenges to tackle.
Design layout, UX (User experience) could be very tedious work: for example, creating a simple, clean, and eye-catching content with high conversation rate can be a great challenge. AI offers to help here as well, with such tools as AttentionInsight, which is developed to show how people react to your design decision. This AI engine works with predetermined eye-tracking database, hence the accuracy of the analysis is very high and designers can easily use this AI solution on day-to-day basis.
It is obvious that designers tend to use such tool for saving time and experimenting with new ideas to work further. This is why it is common to hear that designers are not working with Artificial Intelligence but rather using an Augmented Intelligence which is a half true. The thing is that such tools like VIZCOM or Artbreeder are based on AI engines but the final product is a matter of designer’s choice, taste, and talent.
Copywriting and AI
There are tons of different AI based tools for making text content, even for specific purposes, with an ability to set limitations, ask to find, and automatically add images and videos, but one thing is clear – these AI tools also serve mostly as a time saver, quick help, fast content draft idea. Below there are some illustrative examples what these tools really have to offer.
Tool AI-Writer can generate articles or reword text, and this is quite self-explanatory. After making some tests with AI-Writer it is clear that initial results are not very useful. On the other hand, some tweaks and trials bring quite good outcome which is usable. There are a lot of such tools, and they can be useful.
For example, Wordai could be good choice for making different versions of actually the same text for e-shop, which has, for example, a product with just colour variations or bottle sizes. This unique content could increase SEO performance, be more likable from a consumer point of view.
Another – Writesonic – has to offer 20+ languages and generate some content for different types of advertising, like Facebook Ads, Product Descriptions, Landing Pages and other. This could be useful for generating various ad copy options, short and simple social media posts. Also, Copy.ai or Jasper is really good with Product Descriptions. If you need something more academic, then ArticleForge can generate some reasonable, monotonic, didactical tone, so students making last night’s paper can be saved, which is still immoral thing to do.
If you already have a text but need serious grammar check – Grammarly can be beneficial AI engine for reconsidering some clumsy expressions, mistakes, and logical imperfections. Free AI based translation tool – Deepl – works just fine. All in all, AI in this field can really be useful and make some product fast descriptions, high quality texts, small blog posts, re-phrase what you already have written or to correct grammar. All of that is about saving some time which is the most important thing nowadays.
Music and AI
Artificial intelligence is a tool that can even help in art sector where not only some imagination, talent but some strict rules, like in music, are applied. Actually, rules are always better for AI, because rules of music composition, harmony, tempo, and other things can be translated into mathematics which is an easy piece of cake for a computer. Maybe that is why music industry gets more and more working AI solutions.
For example, Boomy is quite powerful for those who want to create (rather – generate) some electronic dance, meditation, nice beats, Lo-Fi or even experimental piece of sound which makes sense. It sounds a bit crazy, having a fact that with Boomy you can create music for free of charge, upload to social media like Spotify and earn money from your audience. Release albums, share, edit, add your or anybody’s voice-over and voilà – you just generated real songs…with no ability to create music or understanding of notes. Indeed, AI can make it happen quite easily.
We can find intriguing examples how AI and music could be an interesting pair. A group of software engineers and musicians have won an informal AI ‘Eurovision Song Contest’, employing an artificial intelligence prepared on noises made by koalas and kookaburras.
AI tools can help in making very simple yet much appreciated tasks like music for drone’s footage of your real estate building for sale, prepare some sample for e-commerce or simply just generating some sounds for your Youtube video.
Also, there are huge games, movies, and radio industries where AI, for example, can search and scan all music on the internet to find some copies, unauthorized usage or identify music, which sounds “very familiar” to someone’s copyrighted piece of art. Imagine a music publisher who rules tens of thousands of music pieces and constantly have to scan the internet to make sure everything goes as it was planned. These AI “similarity checks” is an efficient solution for such problems.
Landr solves another problem of sound quality – mastering. To put it simple, if you want to produce catchy tunes you have to tweak your music, its sound to some levels which commonly is taken as “sounds good” example. Mastering could be very expensive and time-consuming jobs. AI can rephrase your music to sound better or to simply meet some strict quality criteria as, for example, Spotify asks to follow.
Influencers on Youtube, Twitch, Tik Tok and Instagram are getting more and more on such tools like AIVA. AIVA offers a powerful tool, perform intense production, and copyright ownership. Why do you have to spend hours thinking what musical background to put on your every post or video, if you can simply generate nice beets in minutes and releasing your content as soon as possible. Time and speed matters! What about quality? It depends on who you ask.
AI is like a child: carefully teach and soon you can expect some results. Artificial intelligence is not completely changing the word and people are not losing jobs because of it. It is quite the opposite – design, art, music, digital marketing sectors are benefiting from these technological improvements.
AI is getting better and better, therefore a real human creativity and ability to follow new technological possibilities can just enrich, expand, and make jobs and processes more efficient. It is more about not spending a whole day in technicalities which can be done with the help of AI in minutes, but rather use imagination and combination of all aspects for the best possible outcome to suggest for a market. AI won’t end human art, but rather spur a new golden age of creativity, so future looks optimistic and intriguing.