Gen AI – A Genie in the bottle or the key to a Pandora’s Box

By L Subramanyan

When Apple released its latest ‘Crush’ ad in May this year featuring a giant hydraulic press crushing creative instruments – Guitar, trumpet, video games, sculpture with paint gushing out in many colors to create an ultra thin, it had to be pulled out within just a few days by the company due to backlash it received. 

Techcrunch called it disgusting while actor Hugh Grant labelled it as the ‘destruction of human experience’. The company apologized and all was well.

Or was it?

It is not that the ad was badly produced. In fact, it had the desired effect – shock and awe. It is also not the first time that Apple had broken the mould to create an ad that was a trendsetter. Remember the ‘1984’ Apple ad? So what went wrong?

In a world which is already beset with massive questions related to Artificial Intelligence, particularly in the last 18 months with the emergence of Gen AI, where issues related to creativity, human intervention, human intelligence are largely unanswered, the Apple ad appeared to strike at the very root of this fear. PR Week quotes Nick Kalm, founder and CEO of Reputation Partners, saying  “creatives were unhappy with the ad because Apple was saying -We are going to crush you and your careers and we are going to make billions of dollars off of it. Have a nice day.” That technology will crush human ingenuity, creativity and all those that we cherish as core to our values. That it was an ad for an Apple product was missed.

And therein lies the problem

It is nobody’s argument that Gen AI in all its Avatars – Chat GPT, Gemini, Claude and others being developed by Amazon et al have led to an enormous amount of enthusiasm and hope amongst people and businesses. Gen AI is being touted not just as the ‘Next Best Thing’ but in many conversations as the ‘First Best Thing’ after possibly the invention of the wheel. 

Some businesses I have talked to are looking at a 90% plus increase in productivity and accuracy. Many others are talking about revolutionizing key business functions – customer care, coding, product engineering, content creation, video creation, HR and employee productivity. We are being promised a new ‘golden age’ of soaring employee productivity, a complete PHD thesis in one night and dizzying growth rates in businesses.

It is surely a mother of all hypes

I thought this would probably be a good time to present a contrarian view of Gen AI and where we could also, as Apple vice president of market communications Tor Myhren rued “missed the mark”.

So here are some of the myths and misconceptions that will get busted in the next few months;

• It is always the context, stupid: No matter how we develop the Gen AI models, content creation with Gen AI will always be absent of the context. Few Gen AI models have the roadmap to build the context, not because of technological handicaps but of the sheer absence of the knowledge of the same. As someone who has spent over three decades in communication, I find that every single piece of communication will simply not work unless the context is built into it. Now imagine this in a B2B marketing scenario. How will your content pieces – Thought Leadership blogs, product videos and everything in between work when the Gen AI model that you are using does not have enough (if any) information aout your company’s specific markets, products, solutions, audiences, buyer personas and so on. Building a content bank without any or all of these is creating flat pieces of artefacts that have no intrinsic or extrinsic value, except short-lived bragging rights on Friday night outs “we use Gen AI for all our content”, and belch.
• Factual errors: One of my colleagues told me an interesting story. When a user queried Chat GPT as to ‘how to cure a rattlesnake bite?, all the answers generated by the tools were the ones you are NOT supposed to do like putting ice, cutting the wound open etc.. As absurd as it may sound, as a content marketer, it has now become extremely easy for us at the hiring stage to identify the resumes that have been generated using Gen AI and the ones that are written by the candidate. There is an automatic rejection process for the former, which by the way is almost two-thirds of the applications and that too to in a content marketing company. 
• Empathy and Tone of Voice: Notwithstanding anything that my friends in the pro-Gen AI tent say, it will be a challenge for a model fed on trillions of petabytes of data to generate empathy with your customer in Meerut as against your customer in Madurai. Last week, I was talking to one of the experts in retail tech when he explained the concept of ‘segment of one’ meaning creating customer segments of each and every customer for hyper-personalisation using Gen AI. Even there, the emotive component of reaching out to that one customer is beyond forbidding from the tech perspective.
• Prohibitive Costs: Creating these Gen AI models is not cheap. According to a report by iTrex group, the total cost of setting up and operating a generative AI solution would include Initial deployment expenses of approximately $80,000 to $190,000 (including hardware, development, and data preparation costs) and Recurring expenses: $5,000 to $15,000 (maintenance and ongoing costs). If you think you can simply put queries to Chat GPT and get your blogs out for free ad infinitum, you need to reassess your strategy.
• Talent: This is going to be a critical issue. Current estimates peg the cost of a fully trained engineer to be around USD 70,000 to 200,000 per year. Another Mckinsey report also suggested that the early adopters and heavy users of Gen AI in organizations have the highest propensity to leave, given their newly found skills.
• Watermarking: In terms of usability, this could potentially become the biggest roadblock for commercial usage of Gen AI output. There are already tools available in the market which can identify if the image or the video has been Gen AI generated or not. Imagine running a very expensive advertising campaign with Gen AI generated images only for your audiences to be told that all the images are Gen AI generated. While the reputational cost of this needs to be estimated, it is fairly clear that  large enterprises will not risk being called out for laziness.

It will be imperative for organizations which are considering implementing Gen AI to at least think of the above issues and factor them in while taking what could be a strategic decision affecting the long term future of the company.