Where Do Companies Fit in an AI-First World?

By Rob T. Lee

When OpenAI dropped ChatGPT on the world in November of last year, it imparted to companies two things: a reality check, and a timer. Artificial intelligence (AI) has existed for years, but never has this technology overtaken the zeitgeist as it did when this sophisticated chatbot blew people’s minds. For years, AI was either a fictional super-smart (and often human-like) construct out of fiction like Star Trek and The Terminator, or a simple mumbling robot paraded across countries for novelty’s sake. 

Let’s face it: the world didn’t really take AI seriously. Not until it was doing homework, coding webpages, and diagnosing (and saving) sick dogs.

However, it quickly became apparent that AI, beyond just large language models (LLMs) and chatbots like ChatGPT and Bard, was a reality, and a true solution for many problems. Companies, especially those that had yet to dabble with this tech, were faced with a reality check: “AI is the new toy on the playground — are we making the most out of it?” More than that, the clock was now ticking: the companies that would be the first to use AI, especially in new and novel ways, would have an edge over their competitors. And so, everybody has been scrambling to use this technology since.

But what does AI truly mean for businesses? 

How corporations are adapting to AI-based technologies and capabilities

The first step, of course, is awareness and understanding. AI is not a one-size-fits-all tool that can be simply slotted into any business model. To use AI, organizations first need to understand what it is capable of, and its current capabilities and use cases. Are they looking for generative AI tools like ChatGPT and Midjourney? Perhaps they’re more interested in models that are capable of machine learning, processing large data sets, and outputting new and improved outcomes. Or maybe computer vision software is a better fit.  

Let’s take Amazon as an example. The massive online marketplace uses computer vision tech in its novel convenience store concept Amazon Go. The retail model is simple but intriguing: A customer walks into the store, grabs what items they need, and leaves, no checkout involved. Essentially, the store is fitted with several cameras, which use computer vision AI technology to track what items a person picks up, and then automatically charges them for those items through the Amazon Go app on their phone. It’s a perfect example of using AI to think outside the box and disrupt a traditional business model. 

Facebook is a data beast. What may appear as a simple platform for socializing and sharing content hides a lot of intricate machinations behind that shiny UI. Using a deep learning AI technology called DeepText, the Facebook algorithm is able to understand exactly what you’re saying, pushes your content to people who it knows are interested in the topic of discussion, and bumps up comments on a post by relevance to the topic. Such is the immense power of the Facebook algorithm (and deep learning AI by extension), one that has infamously brought upon it the ire of the public and government in the past. 

On a more positive note, we have Alibaba, the massive online e-tailer that’s devised a truly smart marketing technique that was ahead of its time. Using natural language processing (NLP), the company had made it possible for merchants to generate descriptions for their products, years before ChatGPT even hit the scene.

Startups, SMEs, and the accessibility of innovation

On the other end of the spectrum, startups and small businesses have also been finding interesting ways to make the most of AI. From using LLMs to generate copy, to improving text fidelity with tools like Grammarly and creating revenue prediction models, AI has established itself as a technology that is as democratic as it is innovative.

Still, with everybody hopping on board this hype train, it is those with the best ideas that will win. When smartphone penetration was rapidly picking up, and mobile networks were improving, a company like Uber saw an opportunity to revolutionize the transport industry, while others were pumping out apps like Angry Birds and Candy Crush. Everyone will be using AI soon. The question is how do you use it in a way that no one has before, in a manner that maintains creativity and relevance? Those who will be able to answer this question will be the winners of this innovation era.

Precautions and misgivings

As a cybersecurity specialist, it would be irresponsible of me to sing the praises of artificial intelligence without equally highlighting its misgivings and threat of misuse.

Data is the lifeblood of artificial intelligence, and naturally, that comes with its own risks and pitfalls, especially at a time when everyone is clamoring for this tech in a fervor of FOMO, often sidelining security considerations. 

So what are the threats that come along with AI, and how can companies big and small safeguard themselves?

Let’s first address the elephant in the room: data privacy. 

When a company like Grammarly reviews a user’s writing and suggests language improvements, what goes on behind the scenes? Have we stopped to consider what happens to our writing? All the text we write on programs where Grammarly is active is eventually stored on company servers and fed into its AI to train it. Every email, article, essay and piece of research a person has written could hypothetically be at risk.

Grammarly’s Privacy Policy offers some peace of mind: “When it comes to user content, by default, we process user content associated with your account to provide services to you (for example, writing suggestions). After that, if you did not choose to retain that content, we either delete or de-identify it.” 

This seems to fall in line with current practices, where end-user license agreements (EULAs) are being updated to reflect the implications of this new tech on user data.

While AI is, well, ‘artificial’ intelligence, let’s not forget that its knowledge and wisdom are derived from humanity, so there’s always the risk that AI algorithms and programs could inherit our biases, or provide skewed guidance. Imagine if there was an AI that was trained solely on newspapers from late 1800s America. It would most likely be racist, sexist, and would push cigarettes as a healthy lifestyle choice, for starters. You are what you eat, even if you’re a smart computer algorithm. 

Even if AI is the big shot of the tech world, we tend to put too much stock in its fidelity, when in fact, these models are prone to failures and errors, and even attacks from third parties. For example, AI models and algorithms can be reverse-engineered, leading to intellectual property theft, and massive digital and physical devastation on the higher end of the scale. With this tech eventually making its way into cars, traffic systems, nuclear plants, and much more, one dreads to think of the consequences of a faulty or hacked AI model.

So, in the race to assimilate AI into their operations, businesses will need to balance innovation and out-of-the-box thinking with responsible safety and privacy considerations.