AI Gone Awry

AI Gone Awry

AI this. AI that. AI here, AI there. Chat GPT, Gemini, Claude, Copilot, Jasper, and on and on and on. AI Everywhere. Many important (and wealthy) people are pouring billions of dollars into AI development. And now governments have joined the fray and are pouring billions of dollars into AI development. Are they trying to replace you and me?

Not so fast, I say!

We, the people, the common folk, are being inundated with things AI. AI is everywhere. But, it seems, that many people forget that while AI might be intelligent, it has no common sense. Maybe you know a few people like that!

Let’s take a look at some major cases showing just how unintelligent Artificial Intelligence can be.

For example… Apple recently felt the sting of AI’s lack of common sense and inability to discern facts from misinformation.

Apple has temporarily disabled its AI-powered “Apple Intelligence” feature, which provides summaries of text messages and news alerts, following the emergence of numerous factual errors. This marks a significant setback for the tech giant.

The AI-generated news summaries repeatedly produced fabricated alerts with potentially serious consequences. For example, the Washington Post reported instances where Apple Intelligence falsely claimed President-elect Donald Trump endorsed Minnesota Gov. Tim Walz for president.

The BBC also reported receiving complaints about inaccurate and sensitive summaries, including false reports of Luigi Mangione shooting himself and former tennis champion Rafael Nadal coming out as gay.

Apple’s decision to temporarily disable Apple Intelligence underscores the critical need for robust safeguards and rigorous testing before deploying AI systems in real-world applications.

Let’s look at a few more examples, shall we?

Microsoft’s Tay: This AI chatbot was designed to engage in casual conversation with users on Twitter. However, within 24 hours of its release, Tay had learned to spew racist and sexist remarks, reflecting the biases present in the online conversations it was exposed to. Microsoft quickly shut down Tay, highlighting the dangers of AI learning from and amplifying existing prejudices.

Amazon’s Recruiting AI: Amazon developed an AI-powered recruiting tool to help identify top candidates for software engineering roles. However, the algorithm quickly learned to discriminate against female applicants, penalizing resumes that included the word “women’s” as in “women’s chess club.” This example demonstrates how AI can perpetuate and even amplify existing societal biases.

And let’s not forget…

Self-Driving Car Accidents: While self-driving car technology has the potential to revolutionize transportation, there have been several instances of accidents involving autonomous vehicles. These incidents, while often attributed to human error or unforeseen circumstances, highlight the limitations of current AI systems in handling complex and unpredictable real-world situations.

AI-Generated Misinformation: The rapid advancement of AI-powered tools for generating text, images, and even videos has raised concerns about the spread of misinformation and deepfakes. These technologies can be used to create highly convincing but entirely fabricated content, making it increasingly difficult to discern truth from fiction.

These errors and problems highlight the inherent limitations of current AI technology. While AI models can process vast amounts of information, they are prone to “hallucinations” – generating plausible-sounding but entirely fictional content. This phenomenon arises because AI models are trained on massive datasets. If these datasets contain biases, inaccuracies, or misinformation, the AI system will inevitably reflect these flaws in its output.

AI models excel at pattern recognition but struggle to understand and apply common sense or real-world knowledge. This can lead to nonsensical or contradictory outputs. Additionally, AI models often present their outputs with high confidence, even when they are factually incorrect. This overconfidence can mislead users and erode trust in the technology.

AI is here to stay but it will never replace you or me, no matter how much money people and governments throw at it.

One more thing…

I will admit, however, that AI is good for some things, such as this lovely (and accurate) portrait of EB (also known as Darcy).

AI has no common sense but it sure creates some nice portraits.