Something I've noticed on dozens of AI websites and other web services is a disturbing lack of respect for paying customers.
For many of these services, we—the users—are treated as mere "VISA cards with legs," whose only function is to pay. When we need technical support, request a refund, or simply clarify a question, the scenario is always the same: no direct contact, no ticketing system, not even a visible support email. What they offer, more often than not, is a "Discord" button or a chat with a bot that solves absolutely nothing.
This is utter disrespect. If a technical issue arises—something quite common—the user is trapped in a maze with no way out. It's frustrating and unacceptable.
From personal experience, I've experienced so many disappointments that I now only subscribe to an AI service after testing the product for free for several hours and confirming that there's a real, functional way to contact the support team. I only pay after verifying that support exists—and responds.
A clear example is "Eleven Labs," a well-known service in the industry. However, despite its reputation, it doesn't offer a clear means of contact—only a chatbot. Furthermore, I tested the automatic dubbing service to convert a video from Portuguese to English, and the result was disappointing: an unnatural, artificial voice with a strong Indian accent. Unacceptable, especially for a paid service.
Heygen is another crappy website with no real help. You click "Help Center" and open a page with dozens of pre-made questions and answers. There's no support email or real contact information, and their chat is a "chatbot" that isn't even real-time and doesn't even create a ticket or forward your message to support. You're stuck in an endless loop of messages with a bot that doesn't solve anything.
It's astonishing how so many AI services expect customers to pay for mediocre results, without even offering decent support.
Most AI video generators—such as Hailuo 2, Kling 2, Runway, Google Veo 3, and others—still make many mistakes. The videos produced often have visual flaws, incorrect details, or behaviors that completely defy the laws of physics. At the end of the month, the result is the same: hundreds of credits wasted on videos that end up being useless and eventually deleted.
The same problem occurs with image generators. Platforms like Flux Kontext, DALL·E 3, Nano Banana, and others still often fail to interpret prompts. The AI often fails to understand the request, even when it's something simple.
A clear example: recently, on the renowned Nano Banana, I asked it to alter an image and turn white roses into red—a basic task. The result? The roses remained white. The AI simply couldn't understand or execute something so elementary.
In short: we're paying for mediocre results. Despite the engaging marketing and high prices, the quality still leaves a lot to be desired. This technology is still far from mature, and it will probably take another two or three years to see real progress—which is ironic, considering we're talking about an industry valued at billions of dollars.
Silvio Guerrinha
0 Comments