Everyone Loves a Trend
I’ve been in tech long enough to witness several trends in technology… consolidation of physical servers to VMs, ‘cloud’, and now the AI revolution. Sometimes, it has felt more performative than logical—the initial days of cloud were often ‘lift and shift’ where old monolithic applications were merely moved to IaaS VMs at a hefty cost because companies didn’t want to refactor the code and the old models needed large compute on dedicated VMs. As time has gone on and we’ve embraced containerisation, the logic of cloud-first has begun to make more sense—stateless and resilient microservices allowing companies to rapidly scale up and down whilst managing configurations with code, rather than relying on traditional single points of failure. These new developments have themselves created a whole market of platforms that help manage resources, observe points of failure, produce insights, and validate security concerns in an increasingly hectic continuous deployment of incremental improvements.
That’s why when I’m asked today, ‘tell me how you use AI,’ I don’t just think about buzzwords, and I don’t treat it like magic. Instead, I think about what the purpose is, which is to be a tool. Tools don’t just operate autonomously; they need our guidance—we can’t just throw our brain out the window. Many articles now are suggesting there is a cognitive decline when people rely overly on AI rather than doing the work themselves. And from personal experience, I’ve also seen the results of being insufficiently prepared when AI creates a shiny PowerPoint—people will always ask questions and you need to actually have the background knowledge, not just a copy/pasted word salad of surface facts in bullet form.
Humans Are Key
One of the most relevant films on the role of technology, in my view, is Desk Set (1957, starring Spencer Tracy and Katherin Hepburn), where Hepburn’s character of a librarian is convinced the new-fangled computer is going to put her out of a job, but finally understands that the computer is designed to improve her work and increase her capacity. Every time I hear that AI is going to replace all of us, this is my first point of reference… some tasks that are lower value will be automated, but each evolution of technology generally works hand in hand to allow us to focus on higher value tasks.
Just about every adult who has used a keyboard and mouse at this point has probably written an email or edited a document in a word processor which has a spell check function and perhaps even a grammar checker. Despite these useful tools, it’s unlikely that we will stop teaching school children spelling and grammar, because the tools rely on the user having a basic knowledge of writing!
To circle back to the longer career example: I remember often I would hear the interview question, ‘how do you learn / find answers’. The easy answer was always, ‘I google,’ but leaving it at this would be a sign of immaturity. True research via google has always required critical steps, such as writing a good query that will bring up relevant results, and secondly, an ability and experience to parse the results and understand which ones are truly relevant, which ones are the idiots, and finally which bits are the nuggets of insight that solve your problem.
Practical Examples
This is the true answer to how I use AI: it’s a tool that I can use to improve my efficiency and iterate faster.
For example: if I’m researching information for a prospect to customise a demo, AI tools do a great job of sorting through press releases, documentation, etc. If a customer has a question, using it to get an initial answer based on internal documentation can be a gamechanger, especially if knowledge in an organisation is fragmented. It can also be a great way to make tweaks to grammar or speed up getting information to a French-speaking customer, particularly where I might need a translation of some of the material.
For programming projects, I find it useful when I have a clear idea of what I want to build, or where I might find an open-source project on GitHub that works on another platform or code language, and then have AI create a derivative that works for my use-case. Critically, I’m very proficient at coding in PowerShell, but if I need to write in JavaScript or Python, AI can be a godsend.
Again, in all these examples, it remains necessary to formulate the question well and have the skill to parse or debug the results. Often AI goes too far on the wording of a message or the code it outputs might be so full of bugs that it doesn’t work. For example: AI has on several occasions invented functions that don’t exist in a language or environment I’m using!
I also find that French language translations of technology jargon are rife with confusion as sometimes AI tools are too literal, sometimes the word that I’d use in France is not the ‘official’ one, or confusingly the AI tool might be influenced by a Quebecois neologism.
As a result, knowing how language, grammar, and/or coding work is still essential to success: as the old saying goes, ‘garbage in, garbage out’ !