The buzzword of late everywhere has been AI. Of how it will be transformative, that it will be the next technological revolution, and that it will be all pervasive. So far, it has certainly lived up to that last claim, in that it has been annoyingly thrust into every single conversation and facet of our lives, whether it makes sense or not! The reader may assume that with that statement, and the title of this article, that I have set the tone (of pessimism) for what is to come. Not quite, for indeed there are genuinely useful applications of this technology; but it is undeniable that some of the claims are outlandish at this point in time, and are tipping the balance of opinion towards incredulity. The title hence is merely an expression of fatigue, of being subject to the mention of AI all the time (for the uninitiated, aiyo is an intrinsic South Indian lament, so popular that it is now part of English!).
Interestingly, AI has been around for a long time, before the recent advent of generative language models such as ChatGPT made AI cool again and brought it back into the limelight. Google’s web search is one of the best examples – the predictive options that appear after you start typing something in the search bar is a form of AI. Probably the confusion about what AI means, and how it has been presented to the public until ChatGPT came along are also reasons for the hitherto low profile. The most intuitive meaning of AI as the layperson understands it, is a computer or machine that can make cognitive decisions just as a human would – in other words, a like-for-like human replacement, if not better. In technical circles, this is termed AGI or Artificial General Intelligence when the machine’s capabilities equal that of a human, and is termed ASI or Artificial Superintelligence when capabilities exceed a human’s. Both of these are yet to be achieved, and the fact that these feature in popular culture as antagonists (HAL 9000 in 2001: A Space Odyssey, Skynet in the Terminator series, and the Agents in The Matrix) has created a perception that intelligent machines will take over the earth subjugating humans in the process. However, many of the more ‘mundane’ applications such as speech, text and facial recognition, chess engines, recommendation systems (when you browse on YouTube, Netflix and Amazon) are examples of AI. In most of these cases, the technology was quietly subsumed into software without much fanfare. In fact, there is a term describing this phenomenon: AI effect, which is described as (quoting from Wikipedia) “where either the definition of AI or the concept of intelligence is adjusted to exclude capabilities that AI systems have mastered”. But having a chatbot speak like a real human, or generate realistic text and images probably gave us the feeling of ‘real’ intelligence, and reignited AI in the public consciousness. It didn’t help that strategy and marketing departments everywhere saw a golden opportunity too.
You would be hard-pressed to find any company that isn’t trying to make use of generative AI in their systems, processes, products or solutions at the present moment. Unfortunately, a lot of these efforts are largely unconvincing, and are only helping to increase AI fatigue rather than improve productivity. For instance, software code templates or snippets are cited as efficiency boosters, but these have proven to work only for simple problems; for anything else, a programmer would need to spend additional time reviewing the AI output, making sure it is free from errors and suitable for use, therefore making it counterproductive. WhatsApp’s Meta AI feature is also a curious addition – it appears in the search bar, as well as a standalone button – you can converse with it, ask questions and so on. Since there is no option to disable the feature, it does seem rather sneaky! I am not exactly sure of what use it is, but given that WhatsApp has become an inextricable part of the digital lives of several millions of people – many of whom use it a sole source of news and information – it can quite possibly have negative and dangerous consequences, such as emotional dependence, misinformation, and social withdrawal. It can certainly be argued that all of these outcomes are possible even without AI, but there is no denying it acts as an accelerant. Indeed, that is one of the worries about AI; since it is only as good as the datasets with which it is trained on, if the source is not free from bias and inaccuracies, there are risks of AI acting incorrectly and unfairly. There are already worrying examples of genuine health insurance claims being denied, and rental applications getting rejected despite the applicant having a good record, due to use of faulty AI models.
There is no denying that AI is going to become an inseparable part of our lives going forward, whether in a visible or invisible way. However, the current hype train is underplaying the threats posed, with rapid deployment taking precedence over exercising caution and without applying the required guardrails. We live in an imperfect society, so there is never going to be a perfect and ideal solution, but solutions need to be carefully thought of so that we can exclaim “AI-yeah” instead of “AI-yo”.
Disclaimer: This is an original piece of work that was written without any generative AI contributions whatsoever. Any textual resemblance to that generated by an AI tool is entirely coincidental.