Biggest AI Failures of 2024
Let's learn from their mistakes
We talk about outstanding AI-related projects here all the time. However, there are always some failures in a young and developing industry.
Today, I invite you to look at the year ending and sympathize (or gloat?) with those who entered the race but failed. What's funny is that the list includes all categories, including gadgets, chatbots, digital influencers, and even medical devices.
And if you're in the mood for something opposite, we've got a post about the AI industry's top deals for November:
Keep your mailbox updated with practical knowledge & key news from the AI industry!
AI Gadgets
Everyone has heard about the failure of the first two AI devices introduced earlier this year. On the one hand, we should look at each case separately, but on the other hand, Humane and Rabbit Inc. have so much in common that they must be reviewed together. Consider this: both are Californian startups, both have developed gadgets, both have built their AI assistants, and both are set to hit the market in the first half of 2024. Oh, and both turned out to be huge disappointments.
If you want to know more about the specs of the Humane AI Pin and Rabbit R1, well... Why? Okay, here's a proper post:
Rabbit R1 was unveiled at CES 2024 as a unique and compact gadget. It offered its own Rabbit OS platform, which was supposed to work as a better analog to Siri or Alexa. The user was supposed to be able to solve any task with it, be it creating reminders, searching for information, or working with documents. AI Pin from Humane echoed this idea in almost everything.
Share this post with friends, especially those interested in AI Insights!
The main difference between them was the absence of a screen. Instead of it, the AI Pin offered a miniature projector that could display information in the palm of the user's hand or on the wall. Humane's founders said their gadget would be called upon to destroy the smartphone.
Meanwhile, the AI Pin itself was noticeably more expensive than the Rabbit R1: $699 (plus subscription) versus $199.
But the “pocket AI” victory was not destined to come true. When these gadgets went on sale, dozens of negative reviews appeared on social networks and the media. Reviewers quickly concluded that both devices were useless. Essentially, they were a pair of old Android smartphones with built-in AI assistants.
The bottom line is that sales for both startups failed.
For now, the companies are gradually reducing the cost of the AI Pin and Rabbit R1 and fixing their mistakes. But you can't change the past: both devs will be remembered as foolish attempts to turn the game around with trending technology.
AI Chatbots
If AI gadgets are the newest category in which startups are screwing up, chatbots are the most popular. Even the biggest brands, including OpenAI, Microsoft, and Google (especially Google), have gotten into silly stories where their models behaved inappropriately. So, picking a winner here is problematic.
Here are a few stories where AI has worked hard to ruin a brand's reputation.
260 McNuggets in Addition to Your BigMac
A couple years ago, McDonald's announced the biggest partnership project with IBM to integrate AI into its fast food restaurant chain. The idea was to use the technology at more than 100 drive-thru cafes. And while this approach generally worked, there has been a lot of evidence over the past year that AI isn't always ready to process orders. Below is an example of how it worked in some cases:
In other videos, the AI offered customers nine servings of iced tea instead of one and thought the ice cream needed bacon.
McDonald's prematurely suspended its partnership with IBM and scrapped the initiative. At the same time, the company noted that it doesn’t rule out future AI use.
Roasting Own Company and Customers
French delivery service DPD also decided to keep up with the trends and used AI chatbots to serve its customers. As in the case of McDonald's, everything went well for a while until the AI encountered Ashley Beauchamp. This customer exposed the issue when he attempted to track a missing parcel and found the chatbot unable to provide helpful information.
Frustrated, he began testing the bot's capabilities, which generated inappropriate content, including responses criticizing DPD's service and other customers. It also started swearing and writing hoku.
The incident quickly went viral on social media, with Beauchamp's post being viewed 800,000 times in 24 hours. In response, DPD immediately shut down the chatbot and began updating their system. The company later started using AI again.
Refer someone who started a learning Journey in AI!
Urban AI Calling for Lawbreaking
What really screwed up big time, however, was MyCity, a chatbot launched to help small business owners in New York City. Powered by Microsoft's Azure, the AI made many mistakes, including calling for lawbreaking. For example, it said bosses could get tips from workers, and landlords could discriminate based on income source. The AI also misrepresented minimum wage laws and the acceptance of Section 8 vouchers.
Despite these problems, Mayor Eric Adams sided with AI, acknowledging its inaccuracies but emphasizing that such problems are part of adopting new technology. The city left the tool online with caveats warning users not to rely on its advice on legal matters. Critics called the approach “reckless,” emphasizing the risks of introducing AI into public services without adequate oversight.
Biggest Cheat in AI Education
But the biggest disappointment in AI this year was Joanna Smith-Griffin, the Harvard graduate and entrepreneur who founded AllHere Education. The company developed an AI chatbot named 'Ed,' which created personalized student learning plans. AllHere gained recognition, earning Smith-Griffin a spot on Forbes' "30 Under 30" and securing a multi-million-dollar contract with the Los Angeles Unified School District.
However, the company's success took a sharp turn in November 2024 when Smith-Griffin was arrested and charged with securities fraud, wire fraud, and identity theft. Prosecutors allege she defrauded investors of nearly $10M by misrepresenting AllHere's revenue and claiming non-existent contracts with major school districts. She is also accused of using company funds for personal expenses.
The repercussions were swift; AllHere furloughed its staff in 2024 and subsequently filed for Chapter 7 bankruptcy. The Los Angeles Unified School District halted its AI project with AllHere, raising concerns about the due diligence required in startup investments and the risks associated with rapidly growing AI companies in education.
Digital Influencers
How many really successful digital Influencers can you name?
Even last year, Meta said that they were the future. In the fall of 2023, the company unveiled its Meta Personas platform. The idea was that popular personalities could use AI to digitize themselves and interact with fans through digital avatars. At the time of the announcement, Kendall Jenner, Snoop Dogg, Tom Brady, and other celebrities were involved in the project. But it turned out to be a no-go for anyone.
The AI version of Snoop Dogg had just 15,000 subscribers (versus 88.4 million for the real one), and Jenner had 179,000 subscribers (versus 292 million). As critics said at the time of release, it was too artificial. In August, Meta confirmed to reporters that Personas was shutting down. Now you won't be able to interact with AI characters embodied by any celebrities.
The company didn’t comment on the decision.
Box Instead of Doctors
The latest fiasco is the least funny of the day but instructive.
Launched in 2017 by several high-profile executives from Google and Uber, Forward set out to revolutionize primary care early on. The startup was developing AI-powered doctor-in-a-boxes called CarePods. These boxes, equipped with ultraviolet lighting and floor-to-ceiling screens, allow patients to visit a doctor without a doctor present, directing them to draw their blood, sequencing DNA, and testing for diseases like COVID-19. All without interacting with other people. The AI had to watch to make sure customers didn't make mistakes.
The company also raised $650M from VC investors and had a $1B valuation.
But this ambitious idea was not destined to reach the market. Since our last case is the most serious, let's analyze Forward's problems. I can point out the following critical shortcomings of the project:
- Technical problems: The CarePods experienced numerous issues, including failures with automated blood draws, canceled lab tests, and patients trapped inside the kiosks.
- High cost: Public records show that each CarePod costs more than $1M, making scaling impossible. Forward planned to launch 3,200 CarePods within a year but could only open five.
- Cash outflow: the company was reportedly set to run out of money in less than a year, leading to significant layoffs as early as the first half of 2024.
In addition, Forward had difficulty raising additional funds in the market. As the startup tried to raise more, investors turned down late-stage investments and healthcare deals, resulting in the death of a rather interesting unicorn.
Final Thoughts
Well, I hope these stories didn't make you think the AI industry is doomed.
Every mistake, from chatbot failures to AI startups shutting down, can serve as a valuable lesson. These moments remind us that while AI is an incredible tool, it still requires our careful stewardship. And if the big brands don't realize this, it's cool that you and I do. It's a competitive advantage.
By the way, we still have some untold stories of AI failures.
Should we do a part two? Tell us in the comments!
Share this edition with your friends!
This article was first published in the Creators AI newsletter. View the original edition.
