Welcome to AI Tool Report!

Monday’s top story: OpenAI’s AI translation tool—Whisper—is producing inaccurate transcripts, full of hallucinations, particularly affecting hospitals.

🌤️ This Morning on AI Tool Report

1. 🏥 Hospitals plagued with OpenAI’s hallucinations?

2. 🌴 How to stay safe online when you travel

3. 🤫 Meet Google’s AI agent: Jarvis

4. 😵‍💫 How to transform document chaos into business clarity with AI

5. 🌳 How to incorporate sustainable practices into an event using ChatGPT

6. 📰 Meta and Reuters partner up!

Read Time: 5 minutes

FACT OF THE DAY

🤔 According to Forbes, AI technology is expected to create 12M more jobs than it’s expected to replace, and 97M AI specialists will be needed by 2025.

STOCK MARKETS

➡️ Follow this link to stay updated with the biggest AI-related moves in the stock market.

 — — — — — — —

TRANSLATION TOOLS

🏥 Hospitals plagued with OpenAI’s hallucinations?

Our Report: Back in September, OpenAI released Whisper—an open-source AI transcription tool that enables transcription in multiple languages—but engineers, researchers, and clinical professionals, using the tool, are growing increasingly concerned over the accuracy of the transcripts it's generating.

🔑 Key Points:

• Over 30,000 clinicians and 40 health systems use Whisper to record and transcribe patient meetings, and although it’s been trained on over 680,000 hours of audio data from the web, many have found it often hallucinates.

• One researcher found hallucinations in 8/10 transcriptions, another found, in over 100hrs of transcripts, half contained misinformation, and a developer found hallucinations in nearly all of the 26,000 transcriptions he created.

• The tool makes up sentences with violent or racist sentiments, has invented diseases, generates nonsensical phrases during silences, and includes common phrases from YouTube, like “Thanks for watching!”

🤔 Why you should care: Although chatbot hallucinations are common (remember Google’s Gemini glue pizza debacle, for example), it’s a little more unusual for a transcription tool to hallucinate as surely it’s just following the audio recording, but in response, OpenAI has thanked researchers for sharing their findings, has stated that they will continue to work on improving “the accuracy” of the model, “including reducing hallucinations,” and has re-emphasized that it shouldn’t be used in “certain high-stakes decision-making contexts.”

 — — — — — — —

TOGETHER WITH SURFSHARK

Use Surfshark to stay safe online when you travel

Surfshark: The industry’s best value, fastest, most secure VPN

Surfshark is an all-in-one VPN provider and security toolkit that hides your identity from websites and secures all your devices when you travel, allowing you to surf the web, in any location, safely.

Surfshark is the industry-leading, affordable VPN that gives you:

• One-click VPN security on all your devices, in over 100 countries so you can use the internet safely, on any device, wherever you are

• Best-in-class, 24/7 privacy features including built-in ad-blockers, data breach alerts, AES-256 encryption, and malware protection, so your data remains secure

• Consistently quick connection speeds so you can game or stream your sports, TV shows, and movies without any limitations

And, at just $1.99p/m, Surfshark is 87% less when compared with other VPN providers.

(Plus AI Tool Report readers get 4 months extra if they purchase a 2-year subscription).

1. Chatnode is an AI chatbot builder that allows users to build and train their own AI assistant for their website or research project ⭐⭐⭐⭐⭐ / 5 (Product Hunt)

2. Beno is for automated customer acquisition

3. Ebsynth animates still images with AI

— — — — — — —

AI AGENTS

🤫 Meet Google’s AI agent: Jarvis

Our Report: Following the release of Anthropic’s upgraded Claude last week—which can take control of a user's computer and perform tasks autonomously, on their behalf—Google is also reportedly working on an AI agent (codenamed Jarvis) that can “take over your web browser,” that could be released as early as December.

🔑 Key Points:

• Jarvis—apparently named after Marvel’s Iron Man’s super-intelligent AI assistant, J.A.R.V.I.S—can complete everyday web-based tasks like conducting research, making purchases, and booking flights.

• Rumored to be powered by Google’s next generation LLM—Gemini 2.0—Jarvis responds to a command by capturing computer screenshots and translating them into tasks like clicking a button or typing text.

• According to insiders, it does operate slowly and needs a few seconds to process and translate the screenshots before completing a task, which is similar to Anthropic’s “cumbersome” Claude.

🤔 Why you should care: Google doesn’t have a great track record when it comes to releasing AI-based products and features, and it’s reportedly considering releasing Jarvis to a small number of paying subscribers first, to help it catch any bugs before it releases it to the wider public, which could push the December deadline out significantly.

— — — — — — —

TOGETHER WITH KOLENA

Transform Document Chaos into Business Clarity with AI

Organizations often struggle to extract meaningful insights from vast amounts of unstructured data such as PDFs, emails, contracts and more.

Join Kolena’s webinar to discover how their AI-powered platform transforms this data chaos into structured, actionable insights.

Kolena enables real-time insights across various industries, empowering better decisions and reducing manual effort.

With adaptive AI-driven analysis, Kolena’s platform unlocks the potential of your data, streamlining workflows and driving business clarity.

— — — — — — —

Monday’s Prompt: How to incorporate sustainable practices into an event using ChatGPT

Type this prompt into ChatGPT:

Results: After typing this prompt, you will get strategies to help you incorporate environmentally sustainable practices in your next event.

P.S. Use the Prompt Engineer GPT by AI Tool report to 10x your prompts.

It’s not too late! Join the AI Reports AI Skill Sprint on Skool and master 6 crucial AI skills in just 6 weeks…

Course 2 (Tomorrow, Tuesday 29th October): How AI Can 10x The Effectiveness of Your Business Newsletters

✅ What You'll Learn: Course 2 is with entrepreneur, Louis Shulman, and he will take you through how and why you should leverage AI to draft, research, plan, analyze, and optimize newsletters, troubleshoot any tech issues, and use it to launch, grow, and maintain your newsletter following.

🫱🏻‍🫲🏻 Connect with Louis here

— — — — — — —

PARTNERSHIPS

📰 Meta and Reuters partner up!

• Meta has struck a “multi-year deal” with news outlet, Reuters, which will enable its AI chatbot (Meta AI) to use Reuters content to answer news-related queries, which is the first deal of this type, Meta has signed.

• Although the terms of the deal remain confidential, Reuters will be compensated for allowing Meta to display its content on platforms which include Facebook, Instagram, WhatsApp, and Messenger.

• This partnership will bring a more meaningful and useful experience to users, and marks a step away from Meta’s usual creative-focused, flashy AI features, like the failed celeb AI chatbots, for example.

We read your emails, comments, and poll replies daily.

Hit reply and tell us what you want more of!

Got a friend who needs to learn more about AI? Sign them up to the AI Tool Report, here.

Until next time, Martin & Liam.

P.S. Don’t forget, you can unsubscribe if you don’t want us to land in your inbox anymore.