At one time, the concept of artificial intelligence was limited to films such as Terminator 2: Judgement Day and 2001: A Space Odyssey. We are now living in an era that is being defined by machine learning, natural language processing (NLP), and proprietary systems such as Chat GPT.
AI has already impacted various facets of our lives, and the online coding industry is no different. What benefits can this technology provide, and are there any potential downsides to consider? The aim of this article is to examine both sides of the spectrum before drawing any conclusions.
Site Design
Although programs such as WordPress are still quite handy for novices, experienced coders have begun to gravitate towards artificial intelligence for the speed at which sites can be created. A website builder enhanced with AI is capable of writing thousands of lines of code within a fraction of the time that would be traditionally required. It can check for errors, provide suggestions, and even perform text-to-code functions. It always pays to work smart as opposed to hard.
Performing Mundane Tasks
Coding is not always a glamorous profession, and the same holds true when referring to related fields such as IT. Individuals will sometimes be required to tackle tasks that are boring, and well below their capabilities. The good news is that AI can once again come to the rescue. From using GPT-4 to modify image file names to bulk data entry, automation here is key. This will also free up professionals so that they can address more pertinent tasks.
The Organic Side of Things
Some coding specialists prefer to mull over a project for a substantial amount of time before the programming ever begins. Others might instead think out loud to collect their thoughts. In such a case, why not employ advanced speech recognition software? These bundles are easy to install, and many basic versions can be downloaded for free. The text files can thereafter be exported in various formats; providing a convenient means to transform a vision into a reality.
Are There Any Drawbacks?
Some experts have nonetheless cited a handful of concerns in relation to AI-assisted coding. One common complaint is that this method fails to foster creativity; perhaps resulting in websites that display a cookie-cutter appearance. Other concerns include:
- Inaccuracy on occasion
- Professionals who become completely reliant upon algorithms
- Security issues
- Possible legal concerns
- Implementation difficulties
Of course, many traditional coding specialists are likewise worried that a day may come when their services are completely replaced by an automated program. Although some would call this nothing more than a reactionary stance, others could cite the fact that such changes have already begun to take place (albeit on a smaller scale at the moment).
Whether or not we like it, there is little doubt that artificial intelligence is here to stay. Coding professionals will therefore have to adapt to what can only be called a constantly changing landscape.
AI is ridiculously old. Every computer playing a game of chess uses AI. This was available for home use as early as the 1980s. OCR software uses AI. Spam filters use AI. It’s a wide field.
What’s “new” and hyped is “large language models”. And while theory and algorithms evolve as well, you shouldn’t overlook the major evolution is the huge amounts of data processed, and still it remains a “probabilistic” approach in the end. It’s impossible to reliably prevent a chat bot based on LLM from telling you complete and utter gibberish.
I did a little experiment with GPT3 quite a while back, I gave it some interface definition (in C) from one of my projects and asked it for a sample implementation (giving a few simple requirements). It worked as designed: It gave me nice-looking code that would have just crashed in practice. Telling it “that can’t be quite right because …”, it quickly replied with something along the lines of “oh, I’m sorry, you’re completely right, here’s the fixed code” (which then seemed sane indeed). Pretty much following human conventions here, because that’s what’s found in the huge amount of training data.
No, AI can’t do your coding work. It can most likely assist you with tasks that are “stupid”, repetitive and easy to manually verify once completed, and that’s about it.