Like a hurricane. This is how the new intelligent code assistants seem to have entered the ICT departments of companies. From GitHub Copilot to DeepCode, passing through CodeGuru or TabNine, in recent months there has been a proliferation of initiatives that, based on natural language, are capable of helping developers write or review their code in a much faster and more efficient way.
Almost all experts recognize that they represent a before and after in this field, improving developer productivity and lightening the burden of the most tedious tasks. But at the same time, they warn that they still have a wide margin for improvement. So we wanted to ask ourselves, at what point are we and what can we expect from these intelligent developments in the near future.
From developing… to orchestrating software
One of the great benefits that AI offers in terms of software development is a spectacular increase in productivity. The automation that it entails for many tasks means that engineering teams can scale their ability to iterate and improve the characteristics of your programs with greater speed.
According to Gartner, as algorithms improve, in the near future programmers will increasingly serve as orchestrators of coding tasks, with code wizards completing the bulk of the work.
On the other hand, this type of assistants have the capacity to considerably reduce the barriers to entry in software development, which means opening the door to new companies and creating a more competitive space that results in greater pressure to innovate. In this sense, it is possible that companies that decide not to incorporate these algorithms may suffer to keep pace in a space that may experience strong tensions in the medium term.
In an environment that will either be smart or not
Although of course the use of these code assistants can mean a before and after in the way developers work, it must also be taken into account that they will have a deep impact in the market of professional tools that are placed in the hands of software engineers.
Companies developing these kinds of products will need to start considering how they integrate code helpers into their solutions if they want to capitalize on the expectations of developers more than willing to embrace tools that relieve them of some of the heavy lifting that comes with them. they are currently facing.
In this sense, development environments (IDE) with code assistants are expected to replace basic code editors and in the medium term, programmers will abandon platforms that do not offer this option to opt instead for competitors that do have a strong commitment to integrate augmented intelligence capabilities in all areas of the development process.
The use of Low-Code/No-Code environments will be democratized (even more)
If low-code/no-code solutions are already having a strong impact in a large number of companies, that impact will be even more profound as they incorporate the underlying technology into intelligent code helpers, especially in application development. outside of classic IT environments.
Gartner predicts in this regard, that by 2025, 80% of custom technology solutions within companies will be created by those who are not full-time technical professionals, compared to 20% in 2020. Advancing in the generative processes and workflows It will be the natural step from the current situation, which starts with task-based code generation.
This will further democratize access to this type of solution, developing more easily the figure of the citizen-programmer and speeding up the training actions to achieve it. If until now these people learned basic tasks that did not always fully enter the world of programming, now they will be able to design and build complete applications that combine front-end and back-end services. In fact, we have already begun to see the first examples of how simply using voice commands there are already people capable of developing basic web applications.
But the risks cannot be ignored.
Despite their many and undeniable advantages, intelligent code assistants are not (nor are they likely to be in the medium term) the answer to all business needs. They still represent a nascent technology, which requires continuous monitoring.
As we have pointed out, these developments will improve the productivity of programmers, but they are still a complement and they are not expected to be able to replace them in the short or medium term. Especially since although they speed up code development, there are risks of intellectual property protection, inherent errors in software development, security vulnerabilities, general code quality, etc. that will have to continue to be monitored.
In this sense, the organizations that use these tools must be vigilant about these issues and clearly determine for what and where exactly it is interesting to use them, taking an active awareness of these risks.
