Esker on Air S6 E6: Everyone Has AI — But Not Everyone’s Using It Well

In just a few years, AI has gone from a futuristic concept to a must-have in the world of business software. ChatGPT is a household name, and nearly every software vendor claims to be “AI-powered.” But as businesses rush to keep up, one question often gets overlooked: Is all AI actually good AI?

The truth is, not all AI is created equal. Some solutions are thoughtfully designed to solve real business problems by improving efficiency, reducing errors and empowering employees.  Others? They're just riding the AI hype train with little to offer. To navigate this sometimes confusing landscape, businesses need to know how to tell the difference between effective AI and ineffective — or even harmful — AI.  

AI is supercharging business processes 

From chatbots to fraud detection, AI has numerous applications. Common ways companies use AI today include:

• Predictive analytics: Forecasting trends and risks (example: predicting seasonal demand for products to optimize inventory) 
• Natural language processing (NLP): Chatbots, virtual assistants and text analysis (example: generative AI like ChatGPT falls under NLP, helping businesses automate customer interactions and content creation) 
• Computer vision: Quality control and image recognition (example: visual product searches or detecting product defects in manufacturing) 
• Recommendation systems: Personalizing user experiences (example: streaming platforms suggesting shows based on your viewing history) 
• Robotic process automation (RPA): Automating repetitive workflows like data entry (example: automating order entry to save time and reduce human error) 

At Esker, we leverage AI to power smarter document processing, detect fraud and streamline workflows — all with a “human-in-the-loop” philosophy that ensures people remain in control. 

Spotting the signs of “bad” AI 

Some solutions incorporate AI just to seem innovative without actually solving real problems. But even a great AI solution can turn bad when a business misuses it. At the end of the day, an AI solution is only as good as its design and implementation. So how do you spot an AI solution or strategy that’s destined to fail?

AI red flags to watch for:

• Lack of transparency: AI should never be a black box. Users need to understand how decisions are made. Not explaining how AI algorithms work to customers or stakeholders leads to mistrust and concerns about privacy. 
• Over-reliance on AI outputs and automation: Replacing human judgment in areas requiring nuance (like hiring, compliance and strategic decision making) is very risky. AI-generated insights should be verified with real-world data. And automating tasks that require human judgment, creativity or complex problem-solving should be approached with extreme caution.  
• Data bias: If the data used to train the AI is flawed, so are the outcomes. Failing to identify and address biases present in the training data can lead to unfair or discriminatory outcomes. 
• Poor implementation: Not adequately training staff on how to use AI tools effectively can lead to confusion and misuse. 
• Privacy issues: Data should be collected ethically and used responsibly. Collecting and using customer data without proper consent or exceeding necessary data collection practices is a huge no-no. Data should be collected ethically and used responsibly. Collecting and using customer data without proper consent or exceeding necessary data collection practices is a huge no-no. 
• Outdated models: AI must evolve alongside your business to stay useful. Regularly reviewing and updating AI algorithms ensures they remain accurate and relevant in changing environments. 

In short, bad AI often results from misuse, overconfidence or a lack of oversight. 

What makes AI “good”? 

The best AI doesn’t try to do everything. Instead, it works quietly in the background, empowering users rather than replacing them. Here’s what businesses should look for in a well-designed AI solution and strategy: 

• Solves a specific business problem — good AI starts with having a clear purpose 
• Is trained on high-quality, relevant data that can be monitored and adjusted over time 
• Offers transparency and explainability 
• Keeps humans in the loop — lets AI do the heavy lifting while people handle the exceptions 
• Includes strong governance and regular updates 

At Esker, we build our AI to assist people, not make decisions for them. It combines the speed and consistency of automation with the empathy and insight of a human workforce. 

The bottom line: AI should augment, not replace 

AI is great at handling large volumes of data, spotting patterns, automating repeatable tasks and even drafting emails. But it’s not a magic wand. Human context, creativity and judgment are still irreplaceable. 

The businesses that succeed with AI will be the ones that understand its strengths AND its limits. They’ll use it to eliminate inefficiencies, not to cut corners. And most importantly, they’ll make sure their AI is designed to support their people, not sideline them. 

Want to see what responsible, effective AI looks like in action? 

Explore how Esker’s AI-powered automation tools help Source-to-Pay and Order-to-Cash teams work smarter, not harder.