The case for a more intelligent approach to artificial intelligence.
By Manoj Chaudhary, CTO
It seems like you can’t go a day without seeing another article about what AI will revolutionize. This week, it ranged from pharmaceutical development and publishing to landmine removal and our sewer systems. Running alongside them are at least as many takes on how AI will ultimately affect humanity, from “saving the world” to “sealing our fate.”
Whether you’re an AI “zoomer” or an AI “doomer,” one thing’s for certain: When it comes to artificial intelligence, all systems are full steam ahead. But while AI has the potential to create positive change on a scale we can scarcely imagine, does the current headlong dash really offer us the best chance of getting there?
I subscribe to a different path: A more incremental, thoughtful and deliberate approach that allows workers a spot at the table. This philosophy produces AI designed as a force multiplier for humans, letting them accomplish more than either they or the AI could on their own.
We All Want to Change the World
Economists have traditionally taken a rosy view of new technologies. The benefits may not be equally applied, they’ve argued, but these developments generally lift all boats, with no one left worse off. But a growing trove of more granular data is challenging these beliefs, resulting in models that predict two diverging paths for an AI-empowered world.
One of these is based on a top-down, rapid-fire approach to AI. This philosophy seeks to produce AI good enough to replace humans, with the ultimate goal of eliminating roles.
In addition to the obvious economic implications, there are other problems with this approach. We’ve already begun seeing the drawbacks of the executive-centered, “ready, fire, aim” approach to AI.
McDonalds just ended a two-year pilot program for its AI order takers. The McAI consistently misunderstood orders, with issues ranging from the chatbot adding bacon to someone’s ice cream, to an order misinterpreted as being for “2,610 Chicken McNuggets.”
Similarly, a judge recently ruled against Air Canada over false information their chatbot gave a passenger. The reasoning, it seemed, was that if AI was going to take on human roles, it also needed to take on the accompanying responsibilities.
You Can Count Me Out
These cases illustrate the main problem with some companies’ approach to AI: By attempting to force a revolution on their timeline, they’re ending up with worse products, a poor customer experience, fewer jobs and greater risk – all for few tangible business benefits.
There’s even a term for this, coined by MIT professor Daron Acemoglu and Boston University’s Pascual Restrepo: “so-so technology.” This technology tends to be just good enough to replace human jobs, but not good enough to meaningfully improve anything.
The classic example is self-checkout technology: It worked well enough to eliminate some, but not all, cashier jobs, but it didn’t actually work well. It irritated customers as well as employees, who found themselves spread thinner than ever while having to both monitor customers and manage the machines.
Now we’re seeing the pendulum swing back in the other direction, as retailers eliminate, limit or sharply scale back their reliance on this technology. Some call this a backlash, others call it a correction, but I see it as an industry recognizing the error of trying to go from zero to revolution without doing any of the steps in between. In other words, they tried to force a revolution, when they should have been seeking an evolution.
You Know It’s Gonna Be Alright
Jitterbit has already been working on the next phase of our product roadmap, and AI figures heavily into that. But we aren’t seeking to “revolutionize” our products. We aren’t going to be falling into the trap of bolting on AI for AI’s sake, and winding up with features that are at best “mildly useful” and at worst, irritating, frightening or pointless.
Social-media giant Meta recently made headlines when it announced that it would begin training its generative AI models on user-supplied data – whether users wanted it or not. It did say it will “review objection requests in accordance with relevant data protection laws,” but that means little in places like the U.S., which have few such protections.
The EU is a different story, however. Meta is now being forced to pause its AI rollout there due to regulatory pressure, attracting even more negative media attention just one year after it was fined $1.3 billion for similar EU privacy violations.
The number and scope of these regulations will grow alongside generative AI–not to mention the growing pains that the AI platforms themselves will face, such as the repeated downtime of OpenAI’s ChatGPT in early June. If your product depends on the offerings of just one company, what will happen if that company is faced with a similar disruption?
As part of our more evolution- and infusion-based approach to AI adoption, we’ve planned for this eventuality by creating technology that is completely AI-agnostic. The beauty of the Large Language Model (LLM) landscape is the sheer number of options available, with each offering its own set of benefits.
We’ve long been committed to bringing together the best of the solutions you love, allowing you to integrate them for greater efficiency and usability. We’re taking the same approach to AI. By recognizing AI’s true value comes from being able to best utilize your own data, the next iteration of Jitterbit developments will allow companies the freedom to work without being dependent on any particular Gen AI platform.
By giving you the power to use a wide range of LLMs for different tasks, from Mistral and Perplexity to Big Tech’s Copilot and Azure AI, you’ll be able to switch seamlessly between tools, with your data – not the particular LLM you select – being the key differentiator.
Over the coming months, we’ll be offering more details on these developments. But in the meantime, know that Jitterbit’s AI evolution will be intentional, it will be meaningful, and it will be greatly beneficial to your workforce and your business.
