Sifting through the hype to unmask AI imposters
- Written by: Prasad Ramakrishnan, CIO and Senior Vice President of IT at Freshworks
This year will be the year we expose shallow AI solutions for what they really are – impostors sporting a fake name badge. Many SaaS offerings have marketed themselves as AI-driven, but in reality, lack true AI-driven capabilities. This deceptive practice has become more prevalent as AI gains popularity and organisations seek to capitalise on the trend. Don’t get me wrong, there are many real players in this game. However, it’s important for IT pros to be able to spot the difference to maintain their integrity when it comes to giving their organisations the tools they need.
Discerning the genuine
71% of IT practitioners now rely on AI to support their workloads. With this growing reliance on AI, they’ve also become more discerning and less easily swayed by claims of AI capabilities. Too many past disappointments with overhyped solutions, (cloud-washing anyone?), and a heightened awareness of AI technologies have led to this.
Recognising the critical role of AI in their operations, IT practitioners now exercise caution and demand concrete evidence before embracing new AI-driven tools. This growing emphasis on AI accountability drives them to streamline apps, ensuring adopted solutions not only enhance productivity but also deliver tangible value.
Decoding chatbots
Take customer support chatbots, for example. Some chatbot offerings may be marketed as AI-driven but are actually shallow in terms of AI capabilities – meaning that the SaaS provider is advertising its chatbots as AI-powered, suggesting that the chatbots can understand and respond to customers with human-like intelligence. Meanwhile, the reality is that chatbots often rely on pre-programmed responses and basic rule-based algorithms rather than true AI technologies like natural language processing or machine learning. While they may provide automation in handling customer queries, they lack advanced AI capabilities for meaningful conversations.
If there's a real need for AI in customer interactions, organisations should carefully look at available options to make sure they're getting genuine AI and not just pre-programmed responses. To determine if a chatbot genuinely uses AI, IT practitioners should initially test it internally and then gradually introduce it to real customers, monitoring its performance. These steps will help the chatbot get better over time, giving customers a better experience.
AI-driven virtual assistant — or not?
Who wouldn’t want an AI-powered virtual assistant that can schedule and manage calendars? It’s surely a great AI tool that will undoubtedly enhance productivity, right? Adopters, beware. While these virtual assistants are often marketed as AI-powered and claim to have the ability to understand natural language commands and autonomously manage a user's schedule, a closer look at the imposter reveals that these virtual assistants typically rely on basic rule-based algorithms and predefined templates to handle scheduling tasks. So, while they may be able to recognise certain keywords and phrases and perform simple scheduling actions, they lack the advanced AI capabilities to understand complex scheduling preferences, adapt to changes, or make decisions on behalf of the user. In the end, users may find themselves manually making adjustments to their schedules.
A real AI-powered virtual assistant adapts and makes smart choices based on what it learns. To uncover if a virtual assistant really uses AI, IT teams should test it out internally first. See if it understands and responds to how your team naturally talks to it, if it is able to handle complicated tasks, and to see if it learns over time by demonstrating smart thinking in its responses.
Unmasking shallow solutions
Finding what’s real and what’s not takes time, but it’s better to do the legwork upfront to save on future time and costs. To unmask shallow AI solutions, IT practitioners must engage in relentless app rationalisation — evaluating and scrutinising the AI tools they adopt, to ensure they genuinely enhance productivity. Phasing out overly complex SaaS add-ons and only adding solutions that genuinely enhance productivity and provide real value to their organisations is the way forward in 2024 and beyond.
By Prasad Ramakrishnan, CIO and Senior Vice President of IT at Freshworks
