Uncertainty seems like the only thing that’s certain these days. AI's unbridled development has led to the rise of Large Language Models like ChatGPT, which have the ability to spit out working code on demand that would take hours for the average junior software engineer to put together. How will this development influence the nature of software engineering, and how can developers of the future equip themselves with relevant skills in response to this rapidly transforming climate?
But in an age where cyber attackers run amok and consumers demand higher levels of personalisation than ever before, can businesses truly rely on automation for their programming needs without jeopardising their business model? The honest answer is: Only time will tell. What current and aspiring software engineers can do, however, is analyse industry trends and devise a plan of action to maximise their chances of not just surviving, but thriving in a world that revolves around automation.
Understanding your competition is key to surviving in any industry. The main advantages that our competitor in question possesses lies in speed; specifically its ability to write and read code.
There’s no way any human being can physically produce code as fast as ChatGPT, but software developers can use this to their advantage. Instead of burning hours hunting down specific lines of erroneous code or churning out repetitive sections that only require minimal tweaks, developers can now move further up the value-chain and work on writing code that delivers greater impact. However, a consequence of this shift is that junior software development roles dedicated to such low level tasks will quickly become redundant.
What AI cannot do with its current capabilities, however, is think autonomously. Even AutoGPT, requires proper direction before it can generate serviceable outputs. Software engineers who are capable of analysing how their code will ultimately impact the user experience and wider business outcomes will therefore have a greater advantage over their technically sound counterparts as the ability to code becomes less of a barrier to bringing ideas into fruition.
To illustrate this point, let us zoom in on the act of building an app. Any modern application comes with multiple moving layers that will likely be difficult to use or have vulnerabilities that cyber attackers can exploit if it were purely built by ChatGPT — especially after a single prompt. A competent and AI-literate software engineer will be capable of breaking down the app’s components into chunks while feeding Large Language Models with appropriate data and examples to achieve their desired outcome.
After all, if every business out there is creating the same type of programmes, then companies that can offer more unique products will stand out by default. And who better to incorporate modifications into AI-generated code than programmers themselves? Software developers with cybersecurity knowledge can also prove their worth as more similarly built code starts to circulate, giving cyber attackers a bounty of homogenous vulnerabilities to exploit.
Software engineers of the future may therefore find themselves acting more like project managers who oversee robots. To perform this role competently, however, they will still need to have the ability to read and analyse the robot’s output to ensure that anything that it produces is safe for launch. Suffice to say, code-literacy will still be a relevant skill even with coding robots around.
All of the human advantages that we’ve mentioned thus far assume that the average software engineer is already reasonably competent. But how can computer science undergraduates and junior software developers break into the industry in this climate?
The bar has certainly risen, and building functioning models of weather forecast or location tracking apps may no longer be sufficient to impress recruiters and hiring managers. Like their seniors, aspiring software developers will have to stand out by combining their creative powers with the new possibilities that AI brings to the table.
What will truly separate good “AI-whisperers” or prompt engineers from the rest is the ability to articulate what they require to Large Language Models. As former head of AI at Tesla Andrej Karpathy puts it, “The hottest new programming language is English”, and for good reason. The ability to organise one’s thoughts and break them down into instructions that an AI model can work through systematically will determine the amount of efficiency and quality gains that enterprises can reap from AI. Job openings offering as much as US$335,000 for prompt engineers give us a clear picture of how much attention business owners are paying to this potential.
But good prompt engineering requires more than a good command of English. Code-literacy aside, specialists who can interact with Large Language Models that are being integrated into plugins and built on industry specific knowledge will also rise in demand. Software engineers who are therefore capable of combining their knowledge from other domains such as geography, healthcare, or quite literally any field under the sun that has demand, will therefore rise head and shoulders above a competitive field that is purely focused on code.
So demand is transitioning from pure coding to prompting. And with the number of job postings containing ‘GPT’ rising by 51% between 2021 and 2022 and only set to climb even further, learning to prompt correctly seems like the most obvious career move.
While online learning portals such as Udemy have already rolled out courses focuses on ChatGPT and its image-generating counterpart Midjourney, it is important to note that prompt engineering is still a new skill that will likely need to be paired with auxiliary skills in order to be utilised to its full potential.
That said, a little bit of experimenting can’t hurt. With the global playing field level, it’s really about who can stumble upon the most effective prompts at this point. The more entrepreneurial programmers out there can lean into automation to scale up freelance efforts or even start their own ventures now that AI has made it possible for the average person to do more with less.
However, everything that we’ve discussed thus far hinges heavily on the assumption that ChatGPT will continue to remain free and accessible. As we know, GPT-4 subscriptions only offer users a limited number of prompts. At the moment, OpenAI charges US$0.03 per 1,000 tokens for prompts and $0.06 per 1,000 tokens for results. This means healthy cash flow and/or the ability to build one’s own Large Language Models will be more or less a prerequisite for companies that are looking to have undisrupted access to the power of AI.
In an ideal world, everyone would have the luxury of devoting time and energy to upskilling without having to worry about next month’s paycheck. But the reality is that AI is advancing at a rate faster than the average person can keep up. In less than a year since GPT3’s release, its successor GPT4 has already passed MBA exams, law exams, and coding interviews for six-figure positions set by tech giants such as Amazon and Google.
Combine this with anticipations of a recession this year and Singapore’s slow economic increase of 0.1% increase in Q1 2023 and it's easy to see why many are feeling anxious. Furthermore, 68% of companies in Singapore have come out to say that there will very likely be further job cuts in 2023. But the silver lining is that opportunities may be sprouting elsewhere as recruitment sites like Jobstreet report a 63% growth in tech job ads. Efforts at preparing the nation’s workforce for an automated future are also underway, with tech conversion programmes such as TeSA setting goals to train up full-time software engineers by 2025.
Companies are well aware of the risks of AI and aren’t jumping at every opportunity to cut costs. Organisations such as Samsung have already suffered the consequences as employees keyed in proprietary information into ChatGPT to expedite their workflow. Such actions will allow competitors and other unauthorised members to gain knowledge on things that they should not be privy to. Therefore, companies will be well advised to study the use of generative tools such as ChatGPT and come up with protection measures against the possible outcomes before blindly weaving them into their workflow.
In addition, Singapore companies are already facing some of the most acute tech talent shortages worldwide, with 84% of them admitting to facing difficulty in finding the right people. Rather than letting go of staff members who have been painstakingly trained up and gained familiarity with company domain knowledge, some business owners are taking a more strategic approach by having their middle management work more closely with junior staff to implement AI automation tools while holding onto talent and mitigating risk.
The full extent of AI’s impact on the global workforce is only just beginning to unfold. Make sure that you’re not caught unaware of what’s coming next by following TTAB on LinkedIn and Facebook for more insights from our technical experts.
Plus, don’t forget to join the TTAB Career Conversation for networking opportunities with industry professionals or drop us a message on Facebook to receive personalised tech career advice.
Share this
That may be the first thing that comes to mind when...
Has artificial intelligence become good...
The job market has never been tighter...