'Not Rectifiable'? Generative AI Is Both Untruthful Technology Liar And $4.4-Trillion Value Add, Experts Say

Zinger Key Points
  • Goldman Sachs says generative AI needs a $200B investment for full realization, but can its truthfulness be guaranteed?
  • Despite AI's economic potential, its inherent 'hallucination' habit poses a real risk.

No artificial intelligence (AI) chatbot exists today that doesn’t suffer from occasional fabrication. That’s a problem plaguing virtually every entity depending on generative AI systems for productivity and document creation.

From psychotherapy to legal briefs, high-stakes operations may be jeopardized by AI’s predisposition to invent information.

What Happened: Anthropic, OpenAI, and other major AI developers acknowledge the challenge in their large language models (LLMs). They assure that efforts are being made to enhance truthfulness.

Still, the timeline for improvements — and the sufficiency of AI models for high-risk activities, like dispensing medical advice — remains indeterminate, according to a Monday Associated Press report.

See Also: AI Is Being Name-Checked By Tech More Than Ever This Earnings Season

"It's not rectifiable," Emily Bender, a linguistics professor at the University of Washington, told the AP. "The technology inherently mismatches the proposed use cases."

That's a major issue, and here’s why: Generative AI has enormous economic potential (look at tech stocks in the last seven months). It could boost global labor productivity by more than 1% a year in the decade following widespread usage, Goldman Sachs economists Joseph Briggs and Devesh Kodnani said in a Tuesday report.

Generative AI could add anywhere from $2.6 trillion to $4.4 trillion in value across the global economy, according to the AP report.

Goldman says private companies are going to need to make about $200 billion in upfront investments in physical, digital, and human capital by 2025 to acquire and implement new technologies and reshape business processes.

Companies are going to have to invest a cool $200 billion to shape how generative AI will work in the office, and the AP reports generative AI will add up to $4.4 trillion in new value to the global economy.

Great, that's all fine and dandy — but here's the problem.

AI lies a lot. It has an inconvenient habit of “hallucinating," or fabricating information.

“They’re really just sort of designed to predict the next word. And so there will be some rate at which the model does that inaccurately,” Anthropic co-founder Daniela Amodei told the AP.

The result is AI-generated content that strays from facts and reality — a not-so-discrete roadblock to the projected $4.4-trillion in added value for the global economy.

“The models are designed to make things up. That’s all they do,” Bender reportedly said.

Benzinga’s Take: The seemingly inherent characteristic of AI poses a real concern for industries relying on the technology’s accuracy.

Even tech optimists like Microsoft co-founder Bill Gates recognize the need for AI models to discern fact from fiction.

Truthful AI is crucial for the successful deployment of AI in various industries, the reputation and trustworthiness of AI as a whole, global economic implications and U.S. national security

Keeping the above in mind, fixing AI's fabrications isn't just a technical challenge — it’s a necessity for grasping AI’s true potential and value.

Read Next: AMD Q2 Earnings Preview: Analysts Expect Transitional Quarter Leading To AI-Powered Q4 Recovery

Photo via Pixabay.

Market News and Data brought to you by Benzinga APIs
Posted In: Large CapNewsGlobalTop StoriesTechAIAnthropicartificial intelligenceBill GatesChatGPTConsumer TechDevesh KodnaniEmily benderGenerative AIJoseph briggsOpenAi
Benzinga simplifies the market for smarter investing

Trade confidently with insights and alerts from analyst ratings, free reports and breaking news that affects the stocks you care about.

Join Now: Free!