GenAI: What Lies Beyond the Hype?
Eighteen months after the launch of ChatGPT, why is it that GenAI hasn’t seen real adoption?)
At one point in time, asking billionaire Elon Musk this question would have elicited a smirk. However, his recent announcement that his startup xAI will open-source Grok, the ChatGPT rival later this week, strikes out that option. So, we are left with the only obvious answer: The fear of losing out is making everyone agree that GenAI is the next best thing after sliced bread.
Ever since that fateful date in November of 2022 when OpenAI announced their smartest large language model based chatbot to the world, everyone has been falling over themselves to follow suit. Both industry leaders and the media waxed eloquent about the transformative power of GenAI across industries and its ramifications around the future of work.
The fear of large scale job loss caused by this brute force of a technology added fuel to the fire – this in spite of experts and rational thinkers time and again showing up the technology for what it was. With the situation calming down a bit, companies are now wondering what they actually got from efforts to make GenAI work for them.
When the hype died down
And most of them appear to be none the wiser around some key questions that arose when alongside the hype. How can a business take advantage of the promised cost savings? How can they substantially gain from GenAI firepower? How can they actually put these paper-based use-cases on to their shop floors and elsewhere?
In fact, a recent Harvard Business Review (HBR) article says enterprises are still not done integrating traditional AI such as rule-based algorithms and machine learning into their operations. They quoted a study to suggest that 70% of large companies were still wondering how to reap the potential benefits of AI.
As if that’s not enough, GenAI is proving more complex. It can write a lengthy article but cannot accomplish basic data entry tasks such as extracting and classifying driver’s license that traditional AI could do. The article says that navigating traditional AI was like steaming through choppy waters with a state-of-the-art but cumbersome vessel. But, GenAI adds more tonnage, power and a more turbulent sea to the entire mix.
The biggest challenge for adoption is the cost
However, the biggest challenge around GenAI adoption lies in the cost implications over the long-term, not to mention the existing and future regulatory framework. Which brings us to Elon Musk again, as he sued OpenAI for deviating from its open-source roots, having first helped Sam Altman to co-found the company a decade ago to counter Google’s dominance.
The authors of the HBR article note that the current GenAI situation was similar to the start of the millennium when businesses set up websites without clearly recognizing the specific roles that the wider internet would play in omnichannel strategies, not to mention apps and smartphones that have broadened the scope.
However, all the hype isn’t without substance
Of course, this doesn’t mean that the hype around GenAI is all sound and fury. Businesses need to first figure out what outcomes they want and to achieve them what products to use. The market is only going to get more crowded as more data-rich corporations enter the fray in order to offer new and likely innovative solutions. The question is do the problems need them?
What we can safely say is that early movers on this front are leading efforts to monopolize this space. Take the case of OpenAI who introduced a tool to create ChatGPT powered apps. There’s Google battling this and Apple has threatened that they’re waiting and watching and would arrive on the scene with a typical bang.
However, given that this generation of coders are loath to monopolies, solution developers would continue to choose from a large enough bouquet of offerings, which then would require other innovators to come up with toolkits to enable users to operate across different GenAI at the same time. Reminded of the iOS and Android battle royale?
What should enterprises do henceforth?
Given these challenges, here is what we think enterprises could consider when it comes to sifting the hype from the reality on the GenAI front:
For starters, they should choose performance and not hype, which was what ChatGPT experiments of yore presented. Writing a script, or a program code and doing so based on smart questioning isn’t really what enterprises require. So, while ChatGPT can handle words and languages, traditional learning models deliver better results with images.
So, the magic is to refrain from embracing a technology without questions only because rivals are doing so. Instead, companies should just dig deeper and understand their business issues that need quick resolution and then find a suitable AI tool based on available options. Better still, use a bevy of options and come up with your own answer as there’s none that knows the problem better.