Recent news surrounding AI like ChatGPT has made a lot of people worried. While a lot of news outlets covered the fact that AI can do a lot of things including doing students' homework for them, there was also another group of people who may have started to worry about the power of such AI: Programmers.
Which AI Can Generate Code?
ChatGPT isn't the only AI that can generate code, though. Here are a few of the AI tools that can be used to create code and may be worth playing around with or trying if you're a programmer looking into the potential of AI to help you with your work.
By doing this, you'll quickly see where AI can help and where it struggles and work out just how safe your job is.
DeepCode
DeepCode is an AI tool that's used to analyse code and suggest improvements, look for bugs, and find security vulnerabilities. DeepCode is compatible with programming languages such as JavaScript, Python, Java, C#, C/C++, Ruby, Go, PHP, TypeScript, and Swift.
Kite
Kite is an AI that suggests code in real-time and can integrate with popular code editors including:
- Visual Studio Code (VS Code)
- PyCharm
- IntelliJ IDEA
- Atom
- Sublime Text
- Vim
- Emacs
Kite will also provide users with contextual documentation and autocompletion as they work to help boost their productivity.
GitHub Copilot
GitHub Copilot is another programming completion tool that's powered by AI. This one was developed in collaboration with OpenAI. Much like Kite, it can provide suggestions, autocompletion, and code.
OpenAI Codex
OpenAI's Codex is a language model that was specifically trained on code. ChatGPT was trained on a diverse range of different text types and isn't really the best tool for the job.
OpenAI's Codex can be integrated with code editors like Visual Studio Code and focuses mainly on writing code more efficiently.
Tabnine
Tabnine is another code completion tool that works with several code editors and IDEs. It was trained on a large codebase and uses machine learning models to suggest code completions according to context and code patterns.
Find out more about what AI is.
Reasons AI Will Replace Programmers
The AI tools being used to create or complete programming is seen by some as just one of the first steps towards replacing programmers with AI.
You don't need to be a programmer to see how powerful AI can be. Just a few minutes with ChatGPT and you'll likely be quite impressed at how well it can write. While AI does fall short in certain areas, there are several reasons that it could replace programmers.
AI Can Automate Repetitive Tasks
One of the most important things AI offers is how good it is at repetitive tasks. Not only do humans hate most repetitive tasks, but an AI also doesn't get tired or worse at it over time.
While humans need to eat, sleep, and do other things to keep themselves alive, AI doesn't have to worry about that and will do something just as well at the start of the day as it will after 1,000,000 attempts. In fact, a lot of AI will likely be way better by the 1,000,000th attempt, especially if it's learning as it goes.
In terms of coding, this makes AI really efficient. For most repetitive tasks, AI seems like the obvious choice and rather than considering it a way for AI to steal jobs from programmers, it could be seen as a way to stop wasting programmers' time with dull repetitive tasks and move them onto tasks that make better use of their human brains.
AI Can Generate Code Rapidly
Not only can AI knock repetitive tasks for six, but it can also create code very quickly. There are some caveats here, but for the code that AI can generate well, they can also generate it very quickly.
When AI gets it wrong, it gets it wrong quite quickly, too. If you give AI something it can do, however, you can expect results back far more quickly than a human programmer could ever manage.
This is particularly excellent if you were planning on just using some generic code anyway and finding it online because an AI could do that and write it for you.
AI Can Be Used to Debug and Optimise Code
AI can write code quickly, but it can also read code quickly. This can be particularly useful if you get an AI to look at code and search for bugs and bottlenecks and then suggest improvements based on best practices and by comparing the code to the vast codebase that it was trained on.
A well-trained AI has pretty much seen it all before so it can quickly compare your code to a lot of other code and show you how to get the best out of it.
This is another repetitive or time-consuming task that could be taken out of the hands of humans and given to AI to deal with.
AI Can Use Natural Language Processing to Make Code
This is probably one of the more worrying reasons why AI might replace programmers. Through natural language processing, people with little or no programming skills can simply ask an AI to make code that does what they want.
Naturally, programmers will likely know the best ways to prompt an AI to make some suitable code, but if anybody can just ask AI to make some code, then we could possibly see certain programmers being replaced by people with limited programming knowledge using AI to write their code for them.
AI is Going to Improve Continuously
Last but not least, AI is only going to get better. Over time, AI improves and while it's probably not going to fully replace programmers, the better AI gets at writing code, the less need there'll be for human programmers.
There may come a clear turning point where economic factors win out and it becomes so cheap to write code using AI that it just won't be a sound economic decision to continue hiring programmers.
This is the worst-case scenario for programmers and while it doesn't look completely set to happen yet, there may be some big shifts in the industry if AI continues to improve on its current trajectory.
Reasons AI Won't Ever Fully Replace Programmers
There are definitely signs that AI could take certain tasks from programmers, but there are also a great number of reasons why you don't have to worry about AI taking every single programming job on the planet.
Creative Problem-Solving Is Still Beyond AI's Reach
Most AI, including very advanced AI, lacks the creativity and critical thinking that most humans are known for.
Certain programming challenges are still far too advanced for an AI to even begin tackling and when the solution requires some creative thinking, it'll still need a human to do it, even if they do use an AI to get some parts of the code done quickly for them.
Humans Understand Context Much Better
At the moment, AI doesn't really understand why it's doing what it does and in certain fields, a human fully understands the reasoning behind certain decisions.
Certain coding solutions work in some instances but won't work in others and context and nuance are still two things that AI needs more time to work on.
Humans Are Much Better at Complex Decisions
Current AI technology won't think about the moral, ethical, or financial implications of its code and for this kind of programming, you'll need a human.
An AI only thinks about the code that it's been told to write, finish, or optimise, but a human can think about the code on a bigger scale.
Certain Projects Require a Lot of Collaboration and Communication
Bigger and more complex projects have multiple teams of individuals working together. An AI can't attend meetings, agree on the direction of the project, or do any of these more “human” things.
In any service role, you may also be aware of how difficult it can be for some clients to put into words exactly what they want and this is something that an AI will likely miss.
There'll Likely Be Resistance to Fully Replacing Programmers with AI
Don't underestimate humans' ability to resist change. Some of us embrace new things, while others will regularly be the source of the pushback.
A lot of programmers will likely resist something that could risk their jobs (at least not without proper regulation) and while many will welcome AI that makes them far more productive than they've ever been, they'll likely be unwilling to have a technology come along that would make them unemployed.
It's going to be very interesting to see how the future of AI looks, not just for programmers, but for everybody likely to see their work and livelihood affected by it.
Enjoyed this article? Leave a rating!
Loading...
