S&P 500—Dow—NASDAQ—Russell 2000—FTSE 100—DAX—CAC 40—Nikkei—Hang Seng—ASX 200—ALEX—ALK—BOH—CPF—CYAN—FHB—HE—MATX—MLP—NVDA—AAPL—GOOGL—GOOG—MSFT—AMZN—META—AVGO—TSLA—BRK.B—WMT—LLY—JPM—V—XOM—JNJ—MA—MU—COST—BAC—ORCL—ABBV—HD—PG—CVX—NFLX—KO—AMD—GE—CAT—PEP—MRK—ADBE—DIS—UNH—CSCO—INTC—CRM—PM—MCD—ACN—TMO—NEE—BMY—DHR—HON—RTX—UPS—TXN—LIN—QCOM—AMGN—SPGI—INTU—COP—LOW—AMAT—BKNG—AXP—DE—LMT—MDT—CB—ADP—GILD—MDLZ—SYK—BLK—C—ADI—REGN—SBUX—NOW—CI—VRTX—ZTS—MMC—PLD—SO—DUK—CMCSA—APD—BSX—BDX—EOG—ICE—ISRG—SLB—LRCX—PGR—USB—SCHW—ELV—ITW—KLAC—WM—EQIX—ETN—TGT—MO—HCA—APTV—BTC—ETH—XRP—USDT—SOL—BNB—USDC—DOGE—ADA—STETH—S&P 500—Dow—NASDAQ—Russell 2000—FTSE 100—DAX—CAC 40—Nikkei—Hang Seng—ASX 200—ALEX—ALK—BOH—CPF—CYAN—FHB—HE—MATX—MLP—NVDA—AAPL—GOOGL—GOOG—MSFT—AMZN—META—AVGO—TSLA—BRK.B—WMT—LLY—JPM—V—XOM—JNJ—MA—MU—COST—BAC—ORCL—ABBV—HD—PG—CVX—NFLX—KO—AMD—GE—CAT—PEP—MRK—ADBE—DIS—UNH—CSCO—INTC—CRM—PM—MCD—ACN—TMO—NEE—BMY—DHR—HON—RTX—UPS—TXN—LIN—QCOM—AMGN—SPGI—INTU—COP—LOW—AMAT—BKNG—AXP—DE—LMT—MDT—CB—ADP—GILD—MDLZ—SYK—BLK—C—ADI—REGN—SBUX—NOW—CI—VRTX—ZTS—MMC—PLD—SO—DUK—CMCSA—APD—BSX—BDX—EOG—ICE—ISRG—SLB—LRCX—PGR—USB—SCHW—ELV—ITW—KLAC—WM—EQIX—ETN—TGT—MO—HCA—APTV—BTC—ETH—XRP—USDT—SOL—BNB—USDC—DOGE—ADA—STETH—

More Reliable AI Tools Could Slash Business Costs and Improve Decision-Making

·4 min read·👀 Watch

Executive Summary

New AI advancements allow large language models to express uncertainty, reducing costly "hallucinations" and enabling more dependable enterprise applications. This shift means businesses can look forward to AI applications that are both more trustworthy and more useful, though widespread adoption may take time.

  • Entrepreneurs & Startups: Gain efficiency and reduce errors in tasks ranging from customer service to content creation.
  • Investors: Assess AI startups more effectively as the technology matures and its limitations become clearer.
  • Small Business Operators: Leverage more reliable AI tools to automate tasks, improve customer interactions, and potentially reduce operational costs.

Watch & Prepare

Medium PriorityNext 6-12 months

Businesses relying on LLMs for critical functions need to understand this improvement to avoid existing limitations, and the 'bootstrapping paradox' suggests adoption will be gradual.

Watch AI platforms for updates implementing 'faithful uncertainty.' If a pilot project shows a significant reduction in AI-generated errors and improved task completion, consider scaling its use to reduce operational costs.

Who's Affected
Entrepreneurs & StartupsInvestorsSmall Business Operators
Ripple Effects
  • →More reliable AI → increased automation potential for routine tasks → reduced demand for entry-level administrative roles → shift in required skills for small business employees.
  • →AI's improved trustworthiness → faster adoption in sectors like customer service and content creation → potential for AI-generated content to dominate search results → increased need for businesses to employ AI-powered SEO and content differentiation strategies.
  • →Development of 'faithful uncertainty' → greater investor confidence in AI startups → increased funding for AI companies → potential for AI technology to outpace regulatory frameworks, creating new compliance challenges for businesses.
Moody portrait of a young man with colorful light projections on his face.
Photo by cottonbro studio

More Reliable AI Tools Could Slash Business Costs and Improve Decision-Making

Recent breakthroughs in artificial intelligence are paving the way for Large Language Models (LLMs) to become significantly more dependable. By introducing "faithful uncertainty," AI models can now express their confidence levels, offering educated guesses rather than fabricating false information. This development is crucial for enterprise applications, promising to reduce errors and increase the utility of AI tools across various business functions.

The Change

Google researchers have developed a method called "faithful uncertainty" for LLMs. This technique aligns the AI's internal confidence with its output, allowing it to hedge its responses when unsure, rather than defaulting to a definitive, potentially incorrect, answer. Instead of a binary "answer or abstain" mode, models can now say "my best guess is..." This allows AI agents to know when they need to consult external tools or search APIs to confirm information. The approach reframes "hallucinations" not as outright errors, but as potentially "confident errors" that are appropriately qualified. This means that if an AI makes a factual mistake but hedges its response appropriately, it's not considered a hallucination, preserving utility without sacrificing trust.

This technology is expected to become more accessible through techniques like prompt engineering for off-the-shelf models. While deeper integration will require advanced training methods, the immediate impact for businesses lies in the increased reliability of AI outputs, making them safer for critical functions. The full effect will unfold over the next 6-12 months as these capabilities are integrated into existing and new AI services.

Who's Affected

  • Entrepreneurs & Startups: Founders and growth-stage companies can integrate more reliable AI into customer support, marketing content generation, and internal process automation. This reduces the risk of AI-generated errors impacting brand reputation or user experience, potentially accelerating development cycles and improving funding pitches as technology risks are mitigated.
  • Investors: Venture capitalists and angel investors will have clearer signals for evaluating AI startups. The reduction in "hallucinations" makes AI less of a black box and more of a predictable tool, allowing for more accurate projections of a company's scalability and operational efficiency. This could lead to more confidence in investing in AI-driven businesses.
  • Small Business Operators: Local businesses, from restaurants to retail shops, can leverage more dependable AI for tasks like scheduling, customer inquiries, and basic data analysis. The improved trustworthiness of AI means these tools can be deployed with less fear of generating incorrect information that could mislead customers or disrupt operations, potentially leading to cost savings in areas like administrative support.

Second-Order Effects

  • Increased adoption of more reliable AI in customer service roles could lead to a shift in customer service skill demands, favoring those who can manage and oversee AI interactions rather than perform routine tasks.
  • As AI tools become more trustworthy and cost-effective, their integration into marketing and content creation could lead to a homogenization of online content, requiring businesses to find new ways to differentiate themselves.
  • The development of "faithful uncertainty" further normalizes AI as a trustworthy decision-support tool, potentially accelerating its adoption in regulated industries where accuracy is paramount, such as healthcare or finance.

What to Do

  • Entrepreneurs & Startups: Begin experimenting with prompt engineering techniques to elicit more nuanced responses from your existing LLM tools. Monitor AI service providers for updates incorporating "faithful uncertainty." Evaluate pilot programs for automated customer service or content generation, focusing on accuracy and reduced error rates.

  • Investors: In your due diligence for AI startups, pay close attention to how they are addressing AI reliability and hallucination mitigation. Look for companies that are proactively integrating or developing solutions around concepts like "faithful uncertainty," as this indicates a more mature understanding of AI deployment challenges.

  • Small Business Operators: Identify specific areas in your operations where AI could automate tasks but has been too risky due to inaccuracies. Research AI tools that explicitly mention improved accuracy or uncertainty-aware responses. Consider small, low-risk pilot projects to test the reliability of these tools for tasks like drafting internal communications or responding to common customer FAQs.

More from us

Related Articles

Utility workers repairing downed power lines amidst damaged infrastructure. Outdoor setting with chaos.
AI & Technology

Hawaii Businesses Face AI Model Outages: Diversify AI Supply Chains Now

·7 min read
Aerial view of Waikiki Beach's pristine shoreline with the bustling Honolulu skyline in the background.
AI & Technology

Agentic AI Can Slash Hawaii Business Operating Costs by Up to 50%—Here's How to Implement It

·7 min read
Extreme close-up of computer code displaying various programming terms and elements.
AI & Technology

Hawaii Businesses Face New AI Agent Security Mandates Amidst Rising Cyber Threats

·7 min read