AI Isn't Foolproof...These 5 Fails Confirm That

Are you trying to integrate AI tools into your workflow but finding that they don't always meet your expectations?

Join our live webinar on June 12th to gain deeper insights into practical AI usage from two seasoned mobile growth experts who are unafraid to discuss their technological setbacks.?

A colleague recently sent me a TikTok by creator Steve Boots, who breaks down the growing problem of ChatGPT use in academia, which he says is ushering in a “golden age of laziness.”

And it’s not just academia—you can’t spend five minutes on the internet anymore without coming across something that is clearly generated completely by AI. Original thoughts, or at least original writing styles, are hard to come by these days.

If you’ve read this newsletter or followed me on LinkedIn for the past few years, you know that I am a proponent of AI adoption. At AppAgent, we’ve successfully implemented quite a few AI tools into our processes to both speed things up and reduce the risk of human error.

But this comes with a caveat—AI isn’t perfect, and blindly integrating it into your workflow is a recipe for disaster.

In anticipation of the webinar Brian Bowman and I are hosting next week, where we’ll explore some major limitations of AI today, here are five #AIFails to brighten your week:

1. AppAgent’s Creative Strategist, Alex, had the great idea to generate a simple visual reference using DALL-e. But unfortunately for her, DALL-e could not comprehend the idea of “inside/outside the circle”.

Her prompts slowly devolved from "draw an imaginary landscape, summertime. a computer cursor clicks on this landscape, and the zone around the cursor becomes dead wintery and colorless" to a frustrated "DEAD INSIDE CIRCLE, LIVE OUTSIDE. STRICT".

2. One ChatGPT-4 user asked for a photo of a room with NO cats in it whatsoever. Spoiler alert, it produced an image of an empty room…with a cat in it.

3. Have you noticed the trend of uncanny-valley-style headshots flooding your LinkedIn feed? Not so foolproof for people of color.?

4. Be cautious when asking Google for nutritional advice.

5. AI-generated avatars still have a long way to go before they’re indistinguishable from humans. First of all, they need to learn how to blink.

Wanna see more AI fails? Join us next week !

Today's Insights ??

Brian X. Chen, author of the NY Times’ weekly Tech Fix column, is calling out the trend of companies delivering half-baked AI features or products:

“In the past, when companies unveiled new tech products like phones, what we were shown — features like new cameras and brighter screens — was what we were getting. With artificial intelligence, companies are giving a preview of a potential future, demonstrating technologies that are being developed and working only in limited, controlled conditions. A mature, reliable product might arrive — or might not.”

Mind-Blowing News ??

• Good news, gamers: tech giant Nvidia is rolling out a new chatbot “that guides you through games and optimizes your PC.”
• OpenAI is bringing power to the people with its ChatGPT Edu , a tool designed to support text and vision reasoning and data analysis.
• Are AI tools as widely adopted as we think? According to a recent survey , the answer is no.

Let’s see what next week will bring, stay tuned (and subscribe)!