Welcome, innovators.

We all know that Large Language Models (LLMs) are turning industries upside down (in a mostly good way). But you must also understand the limitations of today’s LLMs. Only then can you use them to your advantage.

Below, we discuss a recent study—Understanding the Limitations of Mathematical Reasoning in Large Language Models—published by Apple employees…

In today’s newsletter:

• Today’s must-read stories 📰

• The limitations of mathematical reasoning in LLMs

• Prompt of the day ✍️

Today’s must-read stories 📰 

SAP doubles down on AI 🤖

SAP is integrating AI across its enterprise solutions, including its ERP system, with Joule, a generative AI copilot, to enhance business processes and improve efficiency through automation.

Using AI to build collective intelligence 🧠

AI is often viewed as automation vs. augmentation, but by focusing on using AI to build collective intelligence, managers can enhance decision-making and creativity across the organization.

FDA approves AI startup's echo software ♥️

iCardio.ai, an AI startup, received FDA approval for its EchoMeasure software, marking its first approval and laying the foundation for future heart disease detection algorithms.

Together with Speechmatics

Real-Time Transcription in 50+ Languages

Speechmatics' real-time transcription delivers over 90% accuracy with <1-second latency—no compromises.

With 25% fewer errors than their nearest competitor, Microsoft, enjoy the most reliable speech recognition available.

From customer service voice bots to television subtitling and critical healthcare transcriptions, Speechmatics offers unparalleled speed and accuracy in 50+ languages.

Try it for free today!

Video of the Day 📽️ 

Miscellaneous resources

• AI Today Podcast: No-hype, practical, real-world insight into AI.

• The AI Podcast: NVIDIA explores the impact of AI on the world.

• Last Week in AI: Text and audio summaries of the most interesting AI news.

AI tools 🔨 

• Pressmaster: AI-powered PR publishing.

• Super: Create custom websites with Notion.

• Reporfy: Create insightful reports with the help of AI. 

Together with Innovating with AI

💼 Want to become an AI Consultant?

Our friends at Innovating with AI just welcomed 170 new students into The AI Consultancy Project, their new program that trains you to build a business as an AI consultant. Here are some highlights...

• The tools and frameworks to find clients and deliver top-notch services

• A 6-month plan to build a 6-figure AI consulting business

• A chance for AI Tool Report readers to get early access to the next enrollment cycle

Click here to get early access to The AI Consultancy Project

The limitations of mathematical reasoning in LLMs

Recent advancements in LLMs have sparked interest in their ability to perform formal mathematical reasoning. 

However, a new study reveals that despite improvements, LLMs still show significant limitations when solving mathematical problems, particularly when small changes are introduced to questions.

Key takeaways

• The performance of LLMs drops when only numerical values in math questions are altered.

• LLMs struggle with questions that have more clauses, indicating difficulty with increased complexity.

• Results suggest that LLMs rely on pattern matching rather than true logical reasoning.

• A new benchmark, GSM-Symbolic, shows LLMs' reasoning capabilities vary widely with minor changes.

• Even irrelevant information can drastically affect performance, leading to significant drops.

• LLMs exhibit a lack of robustness, especially with questions requiring multiple reasoning steps.

Our thoughts

These findings underscore the fragility of LLMs in formal mathematical reasoning. Despite their power, current models need further advancements before they can reliably solve complex problems beyond simple pattern recognition.

Prompt of the day ✍️ 

Inside sales representative (create a sales call scripts document)

How it works:

The prompt directs the AI to create a customized sales call scripts document using user input and reference materials, iterating based on evaluations, with strict adherence to set rules and feedback processes.

Why use it:

An inside sales representative would use this prompt to create a sales call scripts document that improves their sales calls by providing structured, persuasive scripts, handling objections, and refining techniques through iterative feedback.

Prompt Example:

{"prompt":"Develop a tailored Sales Call Scripts Document aligned with the user's individual needs, drawing insights from the supplied reference materials. Initiate interaction with the user to obtain essential specifics and resolve any ambiguities. Iteratively refine the Sales Call Scripts Document through consistent evaluations using the given evaluationRubric and gather user input to ensure the end product aligns with the users expectations.}

You can find the prompt in its entirety here.

Find Us On Twitter
Find Us On LinkedIn