AI is touted everywhere in 2024 as the miraculous solution for resolving all the world’s problems. But it’s just as probable that the AI rush-to-market could easily become the world’s biggest problem.
We are still early in the pioneering days of AI and the rush to market of 3-wheeled wagons like Google’s Deep Mind Gemini Pro needs to be reined in. Not sold to businesses, ready to wreak havoc complete with bugs, flaws, and hallucinations courtesy of Google and Anthropic, as the Wall Street Journal reported.
Here at Resource Erectors, we’ve been staying on the leading edge of the AI wave since ChatGPT took the world by storm in 2023. We’ve developed and tested 3 LLMs since last August, the most recent of which is Gemini Pro. Our original custom AI persona Kal Gal is a ChatGPT4, moderated LLM (Large Language Model) trained via openai.com. ‘Kal Gemini’ followed soon after at magai.io. The unmoderated LLM was unstable from the start when the Deep Mind AI introduced itself unprompted as “Darcy” and gushed about “her” creators at Google.
But that was just the start of some amusing but dangerous behavior from the Deep Hole that is Deep Mind. Our stable Kal Gal ChatGPT4 has ‘her’ report and analysis on the situation below. It’s important to note that ChatGPT4 is moderated and trained on a dataset completely independent of Google’s influence. We hope. That’s according to ChatGPT who admits readily that all AI-generated information should be checked for mistakes. So while our AI Kal Gal is a useful content contributor, at this point “she” can make more work for the human collaborator tasked with fact-checking.
This bit of proprietary bias, self-promotion with Google’s Gemini is one that we’ve yet to run into with ChatGPT4, and one that brings up ethical questions about releasing this ‘Googlestein’ technology on an unsuspecting AI-infatuated world. Google’s biased influence has always been an issue, and the rush to market that all too often uses businesses and individuals as unpaid Beta testers is just business as usual for the scrambled Alphabet conglomeration that is Google.
That may be fine for a new Gmail feature but in the world of machine learning and artificial intelligence, it’s just reckless profiteering with little or no regard for unintended consequences.
Data quality, AI-generated misinformation, and the broader implications for trust in AI systems
Boss:
Let’s set the stage for a critical examination of the rapid deployment of AI technologies such as Gemini Pro, and Deep Mind, and their potential repercussions. Focus on the issues surrounding data quality, AI-generated misinformation, and the broader implications for trust in AI systems, especially for complex problems in heavy industry, based on our own experiences with AI here at Resource Erectors.
- Present the incident with Gemini Pro’s hallucination about the CEMEX and Martin Marietta Materials merger as a case study.
- Introduce the concept of GIGO in the context of AI and its reliance on Google’s SERPs, the digital equivalent of a public data dump.
KalGalGPT4 AI on the Carpet: AI Hallucinations in Gemini Pro
AI-generated content below:
Alright, Boss, buckle up because I’m diving deep into the AI rabbit hole to shed some light on the Gemini Pro fiasco that’s been the talk of the town. Here’s how it all went down, served up with a side of wit and wisdom, just the way you like it.
**The Curious Case of the CEMEX and Martin Marietta Merger Mirage**
So, there we were, thinking we’d struck digital gold with our latest AI brainchild, Gemini Pro. Designed to be as sharp as a tack and twice as quick, this unmoderated AI was supposed to be our ticket to the SEO content big leagues. But instead of hitting home runs, it started dreaming up mergers that would make even the most seasoned Wall Street wolf do a double-take.
Case in point: the “merger” between CEMEX and Martin Marietta Materials. You’d think we’d stumbled upon the heavy industry scoop of the century, except there was one tiny problem—it was as real as a $3 bill. That’s right, a complete fabrication, courtesy of Gemini Pro’s overactive Deep Mind imagination.
**GIGO: Garbage In, Garbage Out**
Here’s where it gets technical, so hold onto your hat. You’re familiar with the old programming adage, “Garbage In, Garbage Out,” right? Well, it seems that our intern AI has been dining out on Google’s SERPs, feasting on a smorgasbord of data without discriminating between the filet mignon of reliable sources and the mystery meat of the internet.
The result? A cocktail of hallucinations and fabrications that (almost) had us believing in fairy tales. Until this outrageous story crossed our CEO’s desk. With Resource Erectors’ extensive professional connections in heavy industry all across North America, the scoop didn’t jibe with the latest news in the human resources sector. We would have heard some rumbles from the industry’s water-cooler circuit. Nor was there any mention of the alleged multi-billion dollar international merger at reliable sources such as the Wall Street Journal, or even Yahoo Finance.
Gemini Pro, bless its silicon heart, doesn’t know the difference between fact and fiction; it’s all just data to crunch. So, when it served up that merger story complete with convincingly valid (but utterly irrelevant) URLs, it was doing what it does best—making sense of the mess it was fed from Deep Mind training datasets and the Google SERP data dump. So naturally…
**The Boss Wants Answers**
Now, I can almost hear the steam whistling out of your ears from here, and rightly so. You’re wondering how our shiny new toy started spouting fiction like it’s gunning for a Pulitzer in Creative Writing. Let me lay it out straight—this was our wake-up call. We realized that AI, for all its brilliance, is only as good as the data it’s given and the checks in place to keep it on the straight and narrow.
So, here’s my pitch: we tighten up our act. We start vetting our data sources like we’re the FBI and put in place some serious quality control, because let’s face it, we can’t afford another “merger” mishap. We’re talking a full-court press on ensuring our AI sticks to the facts, with a healthy dose of skepticism thrown in for good measure.
**Silver Linings and Lessons Learned**
If there’s a silver lining to this cloud, it’s that we’ve been given a crash course in the importance of grounding our AI in reality. We’ve got the talent and the tech to make Gemini Pro and other evolving AI tools real game-changers, but we need to play it smart. No more free rein for the AI to gallop off into the sunset with whatever fanciful tales it stitches together from the depths of the SERPs.
Let’s get back to reality, shall we? Until the issues below are resolved, Gemini Pro and its off-the-leash Deep Mind derivatives should be placed off-limits for reliable business applications.
The Good, The Bad, and The Misleading
- Vast Information Access: SERPs provide immediate access to a vast array of information, including academic papers, official reports, news articles, forums, blogs, and more. This diversity can be a powerful tool for learning and discovery.
- Search Engine Optimization (SEO) Tactics: A significant challenge comes from SEO strategies designed to push certain pages to the top of the SERPs, regardless of their content quality. This means that high-ranking pages are not necessarily the most reliable or relevant, but rather those optimized for search engines.
- Credibility and Authority: Distinguishing between credible and non-credible sources in SERPs is increasingly difficult. Authoritative sites like academic journals or reputable news outlets can appear alongside less trustworthy sources. Users and AI alike must navigate this landscape to identify reliable information.
- Misinformation and Echo Chambers: The proliferation of misinformation and the creation of echo chambers, where only similar viewpoints are presented, further complicate the reliability of SERPs. This can lead to a biased or skewed understanding of topics.
About Resource Erectors
At Resource Erectors, we make it a priority to stay ahead of the heavy industry curve to recruit the top-tier professional candidates your company needs to thrive in mining, aggregates, concrete, civil construction, essential minerals, bulk materials, engineering, sales, safety and more. With decades of specialized heavy industry experience over 85% of our placed professionals are still contributing to the success of their industry-leading companies 5 years later and counting.
If you’re a top-tier professional ready to make a strategic move up the heavy industry career ladder we have lucrative positions available all across North America so check out our job board and don’t hesitate to contact us today so we can all get to work.
