Virtual Workers and AI-Based Robots: Trends Impacting Business 2023-2024 | DataDrivenInvestor | by Valentin Vasilevsky | February 2024

In 2023, the most significant IT trend will be AI and neural networks, specifically a specific type of these neural networks – large language models (LLM) that will significantly influence the industry. This is evident from the amount of investment that is currently flowing into companies that apply artificial intelligence in some form, especially those that use large language models. This article discusses the main changes in the technology sector and what to expect in the coming years.

As an investor and limited partner in the venture fund Davidovs Venture Collective (DVC), I can say that we invest in such companies in the early stages. Over the past year and a half, we have invested in a number of AI-related companies. This includes 63 startups, 75% of which are based on AI technologies. Specifically, this year, about 90% of the companies we invested in are connected to neural networks in some way.

Now, most of the IT companies, which were previously not associated with AI or did not emphasize it much, have rushed to apply AI in their systems. Big companies are acquiring small AI startups to integrate them into their business. And that’s no accident.

What is happening now is a so-called market disruption, where a new technology completely changes the landscape of doing business in this area. For example, Adobe, the maker of Photoshop, Adobe Premiere, and other media content services, is now actively using artificial intelligence.

Another example is Notion, which has already implemented artificial intelligence in the creation of notes and knowledge bases. Artificial intelligence is therefore a significant trend that both newcomers and larger players in the market are pursuing. Artificial intelligence has gained widespread recognition this year, but it has actually been used in software products for some time. This year has become the year of big language models. And tasks that were previously unsolvable are now solved using Large Language Models, which is also a revolution of sorts.

What happened next, however, was perhaps even more surprising. Neural networks have become very effective in classification tasks such as image recognition. That was one of the biggest breakthroughs this year. When OpenAI added a feature to ChatGPT: you can upload any image and it will describe what is shown in the photo.

Thanks to this, a significant revolution is now underway in robotics. Previously, specific neural networks were required that had to be extensively trained to recognize a large number of different objects. In order for the robot to recognize basic objects such as a table, a chair, a glass, a bottle, etc., a neural network had to be trained for each of these objects, and the training process was quite lengthy. Moreover, in order for the robot to move in space, again complex algorithms were needed to recognize these images.

With the advent of large language models, it is possible to easily describe any algorithm. You can literally give the robot a task like “go to the kitchen and get me a bottle of cold water”. Because it works on a large language model, there’s no need to explain exactly what it’s supposed to do. He understands from the context that the water is probably in the fridge.

Before the advent of large language models, we would have to give him very precise instructions: get to the kitchen, open the refrigerator, take out the water, close the refrigerator, and so on. All of this had to be included in the manual. However, none of this is necessary now thanks to the advent of large language models. And if the robot opens the fridge and doesn’t find water there, it can also search the cupboards. These advanced neural networks understand details from context and independently construct a logic chain.

All this affects the IT world in such ways:

  • First, everyone wants to invest in AI,
  • Second, all companies are trying to quickly apply AI in their development, because otherwise they will be pushed out of the market by new and more agile companies.

The same Adobe, Microsoft, Google and so on invest heavily in AI startups and buy them at high prices. The result is a third component: in the field of AI there is a redistribution of spheres of influence. This means that other neural networks that work on different principles are replaced by large language models.

For example, the classical field of computer vision, which was used for robotics and image recognition, has become somewhat useless. Because the application of the new technology of large language models renders expertise useless.

If the term AI is now widely used, neural networks used to be called machine learning. In a few days I will have a meeting with someone who has been in this field for several years. He was facing a problem because he works with other types of neural networks. And he has concerns about them becoming redundant as large language models advance.

The second trend: based on large language models, autonomous agents or autonomous AI workers will be actively developed. This is a significant direction where we no longer consider the AI ​​service as a tool, but as a digital employee. When you hire a virtual designer and formulate your request. It delivers a finished result: for example, a website design. You will interact with him as a human — such an interactive mode.

This is a very big direction to the point where complete automation of managers is possible. What people are currently working on. In the field of AI — these are virtual managers who will monitor performance, employee parameters and manage the team. All this will happen in the coming years.

I would observe the largest companies in the field. They are OpenAI, Ethrophic, Google Bard and all the big LLMs we currently have. We are still waiting for the biggest breakthroughs in the field of large language models. They are constantly becoming more complex, larger and more sophisticated.

Leave a Comment