If we believed everything we read online (well, in tech blogs and portals, at least), we’d think that artificial intelligence is a massive revolution that’s already changing everything across every industry imaginable. There are countless articles out there talking about how AI is reshaping everything under the sun, from manufacturing to retail, from farms to space exploration.
Even we at BairesDev have touted it as the biggest technological advancement of our time. It’s hard not to, really. When you see real-life uses of such cutting-edge technology, it’s hard not to fall victim to that overwhelming feeling that we are, in fact, in front of a world-shattering phenomenon that will change everything.
Yet, if you take a step back and read these articles more critically, you might become suspicious of their heavy adjectivization. Is the world really jumping into AI in troves or are we just blowing things out of proportion? Are we living a true AI revolution or is that a mere wish?
There are numerous surveys and reports that can clarify that. For instance, an IBM survey found out that only 34% of businesses in the US, EU, and China have deployed AI. A Gartner report discovered that the average number of AI projects in a company is 4. A NewVantage Partners survey of senior executives says that only 37.8% of them have successfully created a data-driven organization.
We could go on and on. But the important thing about those numbers is that the current state of AI in business isn’t precisely a revolution. Rather, it feels like anything ranging from siloed efforts to timid attempts. Under such a light, it’s hard to think of AI as a colossal technology that’s already in full gear. This, obviously, should lead us to consider why that’s the case. If so many people hold AI in such high regard and millions of dollars are poured into it month after month – what’s keeping it from conquering everything?
The Integration Challenge
It could be easy to say that it’s just a money problem. This could explain why Google, Facebook, IBM, and similar enterprises are spearheading the development of artificial intelligence: they all have massive amounts of money invested in AI. However, doing so would be oversimplistic. To make the shift to an AI culture, there’s a complicated process that a vast majority of businesses aren’t remotely prepared to tackle.
There are several elements that come into play in that. First and foremost, let’s look at mismatched expectations surrounding AI. Most business owners and decision-makers expect an easy AI implementation, something akin to picking a smartphone off the shelf, turning it on, and start using it. Of course, the reality of using AI is far from that – as far as the Star Wars galaxy.
The integration of AI-based solutions into an existing digital environment can be a pretty stressful procedure. Think of all the digital applications your company uses on a regular basis. These might include everything from accounting software and CRMs to marketing platforms, social media, and HR-focused tools. For a particular AI solution to work for a business, it needs to insert itself among all these applications and make them work together like an orchestra.
If you have ever had the chance to participate in the integration process of new software into a bigger ecosystem, you are surely aware of the hardships that await you. Linking custom software with legacy applications, proprietary programs with open-source platforms, and offline solutions with mobile apps can be a daunting task.
Thus, a lot of AI projects end up being applied in a very narrow field: chatbots, marketing automation tools, supply chain management algorithms, and so on. By doing so, companies miss out on the huge potential of AI in the overall way in which they do business. Sure, AI can help automate social media postings and payments to suppliers but a true AI revolution necessarily has to go beyond that.
The majority of companies investing in AI today play it safe because they know that venturing off the beaten path can result in a herculean endeavor that will take colossal amounts of time, money, and effort. Thus, for most of them, AI’s potential remains in the theoretical field – and the revolution gets delayed.
The Expertise Challenge
Let’s say that a company doesn’t fear to tread off the beaten path and goes forward with a fully-integrated AI system, one that feeds from data coming from diverse sources and areas within a company. What’s more – let’s pretend that the integration is seamless and smooth. That would unleash AI’s full potential, right? Not quite. There’s another important aspect of working with artificial intelligence and its subsets: the proper training of the AI model.
Though AI is sold under the AIaaS approach, it isn’t something you buy, plug and play. Beyond the integration with your entire digital ecosystem, there’s the issue of preparing the algorithms to work with your specific information and processes. In other words, you need those algorithms to understand your company and how it works, and for that to happen, you need to define what data is relevant, accurate and meaningful for the AI solution.
Thus, a lot of the effort that builds a successful AI strategy within a company involves data extraction, cleansing, and normalization to prepare said data and make it useful for the AI algorithms. Of course, such tasks need data experts that can adjust your information and the AI solution itself to your company’s processes and overall goals. That’s what’s driving the demand for data scientists up and what’s slowing down AI adoption in a lot of companies: the expertise shortage.
The talent issue doesn’t end there. A company might get the data science talent to get its AI strategy going by hiring experts or outsourcing development to a company like BairesDev. But that won’t free it from having to customize the AI solution for its specific needs. The data scientists can help with preparing the data and training the model but only people within the company can help those scientists in tailor-making the algorithms for the business’ specific processes.
In other words, the company employees that will use the AI software need to be a part of the AI implementation. If the AI algorithms are to provide valuable insights, they first need to be validated by the people that actually know how to judge those insights. This means that the best employees (the more experienced and skilled among a company’s workforce) have to spend a considerable amount of time collaborating with the developers to create an algorithm that can actually provide value.
This ends up in an uncomfortable scenario for a lot of businesses. For their AI implementation to work, they need to pull off their best employees from their regular tasks to help in training the AI solution. That can be a blow to the normal operation of any company. However, failing to include those employees in the development process might render the whole AI integration useless, as the insights it might end up providing may not be relevant or applicable.
Lighting the Fuse
Both of those challenges are conspiring against a broader implementation of AI, especially in the small and medium business sector. To have a true AI strategy in place today requires highly-skilled professionals, deep industry know-how, extended periods of time, and money. Of course, most companies don’t have all that, which explains why there isn’t a true explosion in AI adoption.
Does that mean that the AI potential is doomed to be limited to what we have right now? Absolutely not. There’s a lot of power to be leveraged in AI today, even if the scope in which you can apply it is more reduced than what you would want. Investing in those solutions today can serve as the foundation for the bigger AI infrastructure of tomorrow. Seeing that most business owners believe that ignoring AI implementation is not an option, it’s best to think of AI now and slowly start unrolling it in businesses, regardless of the project’s scale.
The challenges will be there, sure. That’s especially true for companies that aren’t intrinsically data-driven or mostly digital, as they will have to spend more time in making sense out of AI in their respective fields. But there’s a far bigger risk in not doing anything at all. Fortunately, there’s still time to catch up.
The AI revolution hasn’t happened yet and, for the looks of it, won’t happen for some time. So it’s better if you start preparing for when the time comes for this tech to finally fulfill the prediction everyone is making – that AI will conquer everything around us. The fuse is already lit.
If you enjoyed this, be sure to check out our other AI articles.
