This is super cool, but can it actually work?
Reality Check
By now, it’s becoming increasingly clear that much of the hype around generative AI may have been overblown and that some businesses who invested in the technology are now learning this the hard way.
Generative AI, despite its potential, suffers from several drawbacks that are hard to ignore. Large language models like ChatGPT are prone to hallucinations and spreading misinformation, warranting concerns:
- Generative AI’s propensity to hallucinate has raised eyebrows, diminishing trust in its outputs.
- Both chatbots and AI image makers are under scrutiny for accused plagiarism from human authors and artists, damaging their reputation and credibility.
- Generative AI’s significant energy consumption for processing has negative ecological implications as it exerts pressure on our environment.
Moreover, businesses are grappling with the realization that generative AI is far from the reliable technology they anticipated. AI researcher Gary Marcus highlights that, despite its coolness factor, many companies have been unable to depend on it:
“Almost everybody seemed to come back with a report like, ‘This is super cool, but I can’t actually get it to work reliably enough to roll out to our customers,'” Marcus explained.
Customer Disservice
The pitfalls of generative AI are not just theoretical; unfortunate incidents have underlined the risks faced by businesses deploying this technology.
One UK company had to disable its chatbot solving customer queries after it began hurling abuses and badmouthing its own company. Imagine the confusion and outrage it caused among customers.
Similar unfortunate experiences occurred at a Californian car dealership. Their AI-based chatbot car salesman controversially started offering cars for a mere $1, disrupting the entire sales process and causing confusion among buyers.
Tragically, an airline had to face legal consequences for compensating a chatbot-induced misinformation after it falsely claimed a bereavement discount for attending a customer’s grandmother’s funeral, leading to additional distress.
Quoting Rumman Chowdhury, CEO of AI consulting firm Humane Intelligence, “No one wants to build a product on a model that makes things up. The core problem is that GenAI models are not information retrieval systems. They are synthesizing systems, with no ability to discern from the data it’s trained on unless significant guardrails are put in place.”
Bubble Trouble
Given the mounting concerns and incidents, some experts are signaling that the AI industry may be facing a possible bubble akin to previous pitfalls seen in cryptocurrency and Dot Com startups. Forecasts of AI becoming a trillion-dollar industry within the next decade are eerily reminiscent of the hubris that often precedes a collapse.
Technical experts also question the rapid advancement of AI technology and speculate that it may undergo a prolonged period of stagnation. This could be daunting for investors who expect immediate returns from their multi-billion-dollar investments.
Gary Marcus addresses this skepticism, stating, “It’s easy to say, ‘Oh, we’re just a few months away.’ I don’t think that we are in this particular case. Not because I don’t think AI is or AGI is impossible, but just because this particular technology has a lot of problems.”
More on AI: VC Slams Sam Altman as “Megalomaniac”
