We use AI technologies almost every day: when Grammarly checks our texts, when Siri makes a call to our family and when we use Google Translate as tourists in China. AI technology became so popular that it led to a question of whether AI is capable to replace humans.
Could software developers, designers, content-makers, writers and other professionals lose their jobs and be replaced by AI? We have discussed this question with our C-level executives who have more than 10 years of working in the IT industry. We also asked the neural networks whether they could replace humans and the full articles generated by the neural networks you could find below (spoiler: our experts and AI came to a common opinion).
But before all that, we shall dot all the "i"s and say that for now unfortunately we have no AI - Artificial Intelligence - as it is. Today we have only artificial neural networks (pseudo-neural networks).
What are Neural Networks?
Artificial neural networks (ANN) are programs that process enormous quantities of information in the form of databases and provide answers leaning on these databases. Responses are chosen according to a coefficient: the higher it is, the higher the probability that it will be proposed by the network.
ANN are in the constant process of training and learning: when they are new, they know nothing even if they have an enormous database. They are trained by remembering the right answers, and by doing this they could provide the best answers next time. Thanks to the deepl learning the ANN make now the big progress.
Neural networks recombine the ready answers and elements from their database, so they are incapable of producing something totally new and inventing anything. Artificial Intelligence, which is also trained as neural networks, is theoretically capable of producing something different than it has in its database. But we have not come across AI now.
What is the ANN Good For?
Now a large amount of the developer’s time is consumed by writing code, and as assistance to get a bit of advice and quick solution, developers use StackOverflow where they could find the ready answers.
But the ANN could do too. Neural network solutions like ChatGPT could help to write code, program or an application faster and even without mistakes, find a solution and use templates. Such instruments are popular for example with junior developers who are looking for a ready piece of code but they lack a deep understanding. So before the juniors and other developers used Stackoverflow for assistance, and now they use neural networks.
Additionally, ANN could totally simplify code writing, and with ready code from the neural network a developer just needs to test the code
Accelerating the Routine and Monotonous Work
As the IT industry emerged to simplify calculations and has accelerated our work with information as neural networks progressively came now simplified the routine and monotonous work inside the IT sphere.
Neural networks are capable of accelerating routine work, for example, they make writing and rewriting articles an easy task. The same is with the IT industry: using neural networks a developer does not need to look for the ready pieces of code at Google and in the library. The developer could just ask a neural network to find a particular piece of code.
The neural network helps to create a faster: in web and mobile design, a lot of time is spent on drawing mockups of apps. Designers use the ANN to get simple pictures and images faster: a web designer produces a project design and asks a network to generate 45 variants of mock-ups for screens and then choose one. A designer is acting now as a project manager who delivers a task to execute. Then the neural network generates a picture from the chosen one according to the variations and additions of the needed theme. Finally, the designer asks a network to produce a picture of high quality. And this part of the work is done!
By speeding up the work, the neural networks make the costs of such routing and monotonous work cheaper. However, even with the fact that neural networks could give advice and provide help for routine work, they could not improve the quality of the code.
More Time for Creative Work
You could think that the neural networks generate “new” consent, but it is not actually new as it is built on already created content. There is an incredible amount of content now that could be used by neural networks to generate content for many-many years ahead. The neural networks could not create, and they could not produce something new from scratch. Only humans can create.
The neural network could not invent anything, and only humans could make scientific discoveries. Although there are cases when the ANN has contributed to new programming languages, at the current stage of neural network development they are not capable of inventing a new programming language.
But neural networks give you the opportunity to do more creative work! The neural networks displace the monotonous work, so professionals have time to do other types of work, a creative one, for example. For example, developers have more time for thinking over the interface of a product which is the most important part of the work, and no neural networks could do it.
Neural networks give space for professionals to develop models, build business processes, execute business tasks and think over solutions, developers could work on the software architecture and functions.
If the neural network will be progressed further, then in 1-2 years it will be capable of producing code of good quality. The work of developpes could change: they will think over how to correctly write a request to the neural network, and then check the code from the neural networks.
And with this the issue of how to write a request to a neural network steps up: the better it is written, the better results you will get. The principle is the following: the more details you give the better response you will get.
Who will be ousted?
As neural networks help to accelerate work by doing simple and monotonous work, specialists whose jobs are linked to such work could be replaced - the writing of simple pieces of code or drawing a sketch. For example, juniors write code and then think over it, while middle specialists first think over the task to find a solution and then code.
Software architects think over the architecture, seniors choose the technology and the structure of development, write the code for the website and application, and juniors do the simple tasks. With the neural networks, you do not need juniors, for example, as the neural networks are doing their job (to write the simple and often used pieces of code). So for a project, you would need not 3, but 2 specialists with one neural network.
Taking into account that the entry level of entering the IT sphere is low, as all the materials and courses available and people could become junior developers in a short period of time, it could be a huge problem for switchers. If they will use only the ready materials and will not try to get deep knowledge and more experience, they could be easily replaced by, for example, ChatGPT.
So, if you do not create and think deliberately, you could be replaced by neural networks.
What does the AI say?
We asked several neural networks to answer the question of whether it could be possible for AI to replace people. So what did say ChatGPT and Copy.ai?
On average, all neural networks do not see that AI is capable to replace programmers for now. The neural networks say that software development work requires the ability to think critically and creatively.
AIs say that neural networks could automate routine tasks, make the work more efficient and effective, improve productivity and decreases costs. “It's important to understand that AI is not a replacement for software engineers. Rather, it's a tool”, the copy.ai writes. However, Copi.ai even cites TechCrunch: “artificial intelligence is already replacing programmers in many industries, including medicine, banking and finance, and legal services”.
But their opinions about the future capabilities of AI differ: copy.ai supposes that “some argue that artificial intelligence will eventually become so advanced that computers won't need human programmers anymore!” and chatGPT has a more conservative opinion: “it is unlikely that AI will completely replace programmers in the near future”.
By responding to the question of what hinders AI to replace humans, chatGDP names “its inability to fully understand and replicate the creative problem-solving skills that are essential for programming”, as well as thinking critically and creatively, understanding of the context in which the code is being written (performance, scalability, and maintainability of the code). Copi.ai supposes that “there's still a lot of work that needs to be done before AI can replace programmers”, as there are “some issues with accuracy, as well as issues with bias (for example: when using machine learning for object recognition)”. Interestingly, the selected neural networks name different reasons why neural networks could not fully replace humans.
However, the texts the AI produced are the summary of the texts that are produced by people on the internet, so their answers could be considered as the generalization of the information that was produced now for this question carried out using the algorithm.
The AI does not conceive such a thing as a text (even its text has the introduction, main and conclusion parts) and thus texts from the AI contain contradictory information, for example, Copi.ai writes “Artificial intelligence (AI) is already replacing programmers. AI is still years away from replacing programmers”.
- The neural networks, that recombine the enormous amount of data, are capable of accelerating work by being useful to do routine and monotonous work and thus endangering professionals who do that type of work (junior developers and designers, car drivers).
- Software developers could not be replaced by neural networks, but projects could be done faster and cheaper.
- Neural networks provide help for people and make their work more efficient. However, the neural networks could not do the creative and strategic work but they provide people with more time for it.