Generative AI. Forecast for 2023 (Part one)

A few words as intro

For the last year and especially half a year I’ve been hanging out with AI, spending hours every day on searches and tests. It even seems to have become my leisure time for the most part (oh my god, what have I come to…).

During this time, I think I’ve picked up the dynamics of what’s going on and armed myself with a whole armful of sources of information and people in this environment. Now I want to share my thoughts on the topic.

General Trends

This year generative AI and services based on it will penetrate deeper into the masses, find new original applications and become a normal part of personal life, leisure, work and business.

Obviosly, the hype of ChatGPT, MidJourney, Dall-E and Stable Diffusion has thundered and continues to echo not only among users.

Large and small existing players in the industry, teams that are just getting ready to enter the market and industry professionals are now very well aware of the prospects and the cost of each day of delay.

This year we will see a crazy battle for the audience in this global redesign of the entire digital space.

Users will see more than once bright new product releases and those who somehow missed the peak of the news madness with ChatGPT and MidJourney last ~December can catch up with them =) Every couple of months the leaderboard will be in for a shakeup. With the goal of knowing all the latest and best, you’ll have to spend a lot of time researching and comparing, and shape your info feeds accordingly.

I think this is the most thankless and turbulent time in history to buy an annual subscription to anything 😅

“Understood, great. But how will this affect existing food chains in the digital environment?” — you may ask.

Visual-generative AI services, photobanks, Instagram, and the like

What trends I see for generative services themselves:

▪️ 100% noticeable improvement in quality and greater efficiency and control over the creation process (reducing the number of iterations with AI to get the desired result)

▪️ It is not clear at all the dynamics of cost of services for users. It is not clear at what point increasing competition, improving quality and usability, the price of computing power required for continuous training and releases of new versions of algorithms, as well as to serve the rapid growth of queries themselves will find their balance.

But I tend to think that the market leaders, having gained an audience, will raise their prices above the current ones. The average price tag on the market for image generation/upscale/inpainting will decrease or not change.

▪️ A user satiated with the new quality original visuals will raise the bar for everything they see in the digital world. This will lead to higher requirements for art itself, advertising, illustrations in articles, the design of websites and so on. Accordingly, those businesses who will not adjust in time, will lose part of the audience.

▪️ Photographers and digital artists will search and find new interesting hybrid formats at the intersection to win the competition or just spontaneously by virtue of their creativity =)

Photobanks

During the year, many different photostocks will follow their own combinations of the following key tactics:

▪️ Increasing the rigor of AI-generated image selection. Not only on the basis of quality of execution, but also on the platform’s saturation of already uploaded similar instances.

▪️ Opening up their own image generators and upscalers on their platforms and including their use of subscriptions, at first for free, later with an increase in the overall subscription price. Built-in generators, if they achieve the proper level of quality, will eventually be provided with preferences when considering uploading to stock, as well as convenient functions that allow to do everything on the fly: generation → enhancement → tagging → uploading to the site

▪️ Lowering the price of downloading a generated image in order to compete with platforms like MidJourney, where you can download anything with access to the entire catalog with basic 10$/mon subscription.

Overall, I see a significant increase in competition to classic photography and digital art from the content flow of AI generators.

In an increased competition all kinds of creators can exist in a commercial sense, but only with better content and the ability to upload a lot of material to stay visible and have sales.

The lower segment of creators with insufficient quality/non-creative works and low upload rate will not only lose a share of income… income will become less regular, more chaotic, and the prospects unclear, which will alienate many.

The vector graphics artists can temporarily breathe easy, but in this segment the hand of AI will start to reach out closer to 2024, fed up with the chaos induced in the raster segment.

As for video makers, everything is quite smooth. I would say you can not worry at all for most of the year, but at the end begin to keep a hand on the pulse and assess the prospect to adapt in time.

At the moment, there are no services capable of generating commercial-grade video.

Instagram, TikTok, Pinterest, etc.

Here the efficiency of generated content will remain high for a long time, and its creation for such platforms is expedient. Much more expedient than in case of stocks (though in the monetization scheme there are 1–2 additional steps depending on the chosen approach).

But why???

▪️ You don’t have to worry about upscaling, incorrect shadows and focus. There are no quality requirements for uploaded content.

▪️ Due to the peculiarities of traffic flows and in particular the recommendation system on such sites. The demand is rubber-stamped.

In the coming months, Tiktok, Reels, Shorts will be flooded with slides made from generated images with music upon it. This channel is super attractive considering the peculiarities of promotion of these content formats, and the complexity of production is minimal and more than justified.

Freelance (Image creation tasks segment)

▪️ Significant, but not dramatic share of the tasks which were earlier given on freelance, will be performed by your own forces with generators. This is mostly related to sketches/concept art of any kind.

▪️ The growing popularity of tasks from the series: “I’ve generated this. Help me make it better, fix here and there”

▪️ The growth of efficiency of freelancers will increase. Some of them, who will make a bet on AI in time optimizing some steps of their work, will get a temporary advantage over their colleagues, but with time the situation will be balanced.

Thank you for your attention!

If you saw a flaw or would like to challenge some point or add something to what’s written, please share in the comments, I will be happy to talk!

This was the first part in a series of articles on the topic. I will continue to post sequels to other areas in the coming days. The next article in the series will be on text and partly on marketing, then there’s a case study on video blogging and audio generation.

Sign up so you don’t miss it 🙂