How Will Artificial Intelligence Change in 2024?

7 min readMar 7, 2024

The year 2023 was a turning point in the evolution of artificial intelligence and its role in society. Generative AI has taken technology out of the shadows and into the center of public space, making it a lightning-fast consumer product. Never before has an experimental prototype been used on such a large scale, which is why we have not been able to fully comprehend what is happening, let alone consider the impact of this radically new technology on our lives. The rapid technological progress of AI — and the rapid and varied response to it — makes predicting the industry's future not something for the faint of heart. Still, let’s try to look into the future to see what changes the world will face in the next 12 months.

Artificial intelligence in 2023

At the end of 2022, OpenAI launched a free web application called ChatGPT. Although it was hard to believe, such a modest release changed everything—by January 2023, ChatGPT became the fastest-growing application in history and gave any network user access to one of the most powerful neural networks.

Then in February, Microsoft and Google unveiled plans to merge chatbots with search engines. And, I must say, the early demos were not very good — the Microsoft Bing chatbot literally went off the rails, giving out nonsense, and Google’s Bard was completely sinful with factual errors in the commercial. However, this genius released from the bottle would not come back, and eventually, it led to a rethinking of the daily interaction of users with the Internet.

So, in a short time, corporations introduced their chatbot-based assistants to the general public, and their abilities were really impressive—AI systems could write texts, generate images, and create reports and entire sets of slides in a matter of seconds. Meta Corporation also did not remain in debt, releasing two models simultaneously to create images of literally anything.

What’s more, today, Google phones with built—in AI allow you to edit photos to an unprecedented degree, replacing sad faces with happy ones and cloudy days with perfect sunsets. And while the initial hype is fading, and the talk that AI will destroy our civilization seems to have remained in 2023, these innovative, intelligent systems have become a symbol of change in literally all industries — from the economy to education.

Remember how one of the major discussions in 2023 about the role of ChatGPT and similar chatbots in education ended — all focused on the fact that students can use AI to cheat, but as the year went on, it became clear that the disability of teachers to teach schoolchildren and students how to interact with chatbots could put them at a disadvantage. And yes, there was no “revolution” in the education system.

Nevertheless, understanding how to work with neural networks is extremely important—the more we know about these intelligent systems, the more opportunities we have.

Neural networks in 2024

So, since the release of ChatGPT, the development of generative artificial intelligence models has continued at a breakneck pace — a new class of AI systems is learning to be multimodal. This means that the data used for training neural networks comes from text sources, such as Wikipedia, YouTube videos, and other audio and visual information sources. All this once again raises one of the main issues related to AI systems — the reliability of information.

Reliability of information

Chatbots generously share fake photos and videos with us (and it is quite difficult to understand that this is fake), and in the future, this problem will worsen, causing more and more harm to individuals, large companies, and even states. All this is happening despite the nascent regulation, which is why many experts predict the emergence of new, previously unseen classes of problems.

Today, AI allows you to automate the creation of fakes—both text and video—which means that more and more content is imitating the truth on the web. According to The Washington Post and NewsGuard, an organization that tracks disinformation, since May 2023, the number of sites posting false articles created by artificial intelligence has increased by more than 1,000%, from 49 sites to more than 600.

Creating larger models

Despite numerous warnings, the development of existing AI systems continues at an accelerated pace. Recall that last year, more than 1,800 technical specialists, including Elon Musk and Steve Woznik, as well as engineers from Amazon, DeepMind, Google, Meta, and Microsoft, signed an open letter demanding to suspend the training of AI more powerful than GPT-4 for at least six months.

The problem is also the total competition for profit, fame, and dominance in the industry, which began with the release of ChatGPT. Such competition, as history has shown more than once, likes to circumvent all sorts of restrictions and attempts at regulation. One way or another, many experts are inclined to believe that we should be prepared for the emergence of more powerful AI and a whole stream of various applications.

So, in December 2023, Google DeepMind announced the latest artificial intelligence model, Gemini Ultra, without disclosing the amount of computing power used to train it. However, according to the estimates of the organization Epoch, which deals with predicting artificial intelligence, the system was trained with the greatest capacity. And yes, Gemini Ultra is about as good as the experts predicted.

In 2024, chatbots are projected to make fewer mistakes, but “from the user’s point of view, little will change, and the general public’s reaction to future AI models will be akin to the reaction to new iPhone models,” OpenAI CEO Sam Altman recently expressed.

Fight for electricity

“In 2024, the demand for electricity will increase significantly,” says Dan Hendricks, executive director of the Center for Artificial Intelligence Security, a nonprofit organization based in San Francisco. This is because about 20% of the world’s data center capacity is currently used for artificial intelligence. This share will likely increase dramatically in 2024 as AI systems are trained and run on ever-increasing computing power.

Companies will increasingly try to strike deals with governments to secure energy supplies. And because oil is one of the easiest ways to power data centers, oil-rich countries in the Middle East, willing to “invest in stranger investments,” will likely become more significant in the global competition for AI supremacy.

Growing gap

The International Telecommunication Union estimates that about 2.6 billion people — about a third of the world’s population — do not have access to the Internet. This digital divide can determine who can benefit from AI.

“There are many inequalities—inequality in education, income inequality, gender inequality. If we add digital inequality, it will be impossible to close the gap,” says Bolor-Erdene Batzengel, a researcher at the University of Oxford and former Vice Minister of Digital Development and Communications of Mongolia.

Even when users in developing countries gain access to AI, it is rarely designed with their needs in mind. “Moreover, algorithms are written by engineers from developed countries,” experts note. However, this problem is not as well covered as others, and it is too early to talk about the consequences of this “digital divide.” However, as more powerful AI systems are created, inequality will grow.

Even more robots

The transition from using many small models to perform various tasks to a single one is inevitable. This is confirmed by multimodal models such as GPT-4 and Gemini from Google DeepMind, which can solve both visual and linguistic problems. Based on this, we can assume that the same thing will happen with robots—why train one to turn pancakes and the other to open doors if you can create one universal multitasking model?

You don’t need to look far for examples — several examples of work in this area appeared in 2023. In June, DeepMind released Robocat (an update to last year’s Gato), which generates its own data through trial and error to learn how to control many different robotic arms (instead of one specific arm).

In October, the company released another universal robot model called RT-X and a large new set of general-purpose training data in collaboration with 33 university laboratories. Other leading research groups, such as RAIL (Robotic Artificial Intelligence and Learning) at the University of California, Berkeley, are studying similar technology.

While the lack of data is a real problem, scientists are developing methods that allow robots to learn better through trial and error. In short, there will be more robots (especially smart ones) every year.

Go to details

The key to staying one step ahead in the changing landscape of artificial intelligence is to embrace new trends and adapt to them. This means that businesses (as well as governments that invest in the industry) that embrace new trends and adapt to them will not only improve their operations but also pave the way for unprecedented growth and innovation.

Also note that the future of AI in 2024 is really promising and includes not only the points described above but also email automation, generative and conversational AI, and robotic technologies. It is safe to say that this year, AI will become even more specific in every sense of the word.

However, as with any other technology, there will be problems. Some are voiced in this article, but what we will encounter is not yet known. In short, the future will show, and progress will not be stopped.