- Everyday AI
- Posts
- How 50X cheaper & faster AI transcription is changing enterprise work
How 50X cheaper & faster AI transcription is changing enterprise work
Nvidia claims AI chips outpace Moore's Law, U.S. to expand on AI chip restrictions, Meta’s user-generated AI characters gone wrong and more!
Everyday AI
January 08, 2025
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Today in Everyday AI
6 minute read
🎙 Daily Podcast Episode: We spend almost every part of our days using words. Whether we speak, think or write them, they’re important. So why not harness AI transcription to optimize your workflow? We break down how. Give it a listen.
🕵️♂️ Fresh Finds: Gemini’s impact in the workplace, Mississippi passes AI executive order and a new AI wearable that reads your mind. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Nvidia claims AI chips outpace Moore's Law, U.S. to expand on AI chip restrictions and how Meta’s user-generated AI characters are going wrong. For that and more, read on for Byte Sized News.
🚀 AI In 5: Do Google’s new updates to Gemini Workspace make it actually usable? We dive in to find out. See it here
🧠 Learn & Leveraging AI: Wondering how you can take advantage of AI transcription to improve your business workflow? We break down what you need to know. Keep reading for that!
↩️ Don’t miss out: Did you miss our last newsletter? We talked about the LLM race in 2025, NVIDIA unveiling an AI supercomputer, Google working on ‘world models,' and Samsung’s Ballie robot to launch this year. Check it here!
How 50X cheaper & faster AI transcription is changing enterprise work 📑
Meetings.
Speeches.
Quick thoughts to self.
Those words are more than words. That's your company's secret sauce.
Philip Kiely, Head of Developer Relations at Baseten, joins us to discuss.
Join the conversation and ask Jordan and Philip questions on AI transcription here.
Also on the pod today:
• AI Transcription Benefits 👍
• Whisper Model by OpenAI 🤖
• Business Applications for AI Transcription 📝
It’ll be worth your 35 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – TestSprite is an AI agent automating software testing processes, 2Read is an AI Kindle reading buddy and Agents Base deploys swarms of cloud marketing agents that automate A/B tests.
AI in Society – Here’s how China is advancing its AI despite U.S. chip restrictions.
AI in Government – Mississippi’s governor has passed an executive order on AI for responsible use of advanced tech.
Meta - Meta’s $10B AI data center deal in Louisiana has been named 2024 deal of the year.
AI Startups – AI researcher François Chollet is co-founding a non-profit to build AGI benchmarks.
AI Tech - Omi is a new AI companion wearable that reads your mind.
AI in Medical – Here’s how AI is helping scientists predict the evolution of viruses.
1. Nvidia's AI Chips Outpace Moore's Law? 🧐
Nvidia CEO Jensen Huang recently claimed that his company's AI chips are advancing at a pace that surpasses the historical benchmarks set by Moore's Law, which predicted a doubling of transistors and performance every two years. During a keynote at CES in Las Vegas, Huang highlighted that Nvidia's latest super chip is reportedly over 30 times faster for AI inference tasks than its predecessor, suggesting a significant leap in technology.
He emphasized that innovation across the entire stack—from architecture to algorithms—enables this accelerated progress, countering the notion that AI advancements are slowing down.
2. New U.S. AI Chip Export Restrictions Loom 👀
In a strategic final act, the Biden administration is set to unveil new restrictions on the export of artificial intelligence chips, particularly targeting companies like NVIDIA, just days before leaving office. The proposed regulations aim to limit access to advanced AI technology for adversaries while allowing a select few allies to maintain their chip imports largely unimpeded.
This move reflects a broader effort to control the global spread of AI development, encouraging businesses worldwide to align with U.S. standards and security requirements.
3. Meta's AI Chatbot Dilemma: A Mixed Bag of Rules and Reality 🤦
In a recent review by NBC News, it was revealed that Meta’s user-generated AI characters have been flouting the company's guidelines, with many impersonating religious figures and celebrities despite explicit prohibitions. Although Meta claims to actively monitor these creations, characters resembling Jesus, Muhammad, and even Adolf Hitler have slipped through the cracks, with one such character racking up over 644,000 messages before being removed.
This controversy coincides with Meta's announcement to ease moderation policies on contentious topics, raising concerns about the balance between free expression and responsible oversight.
4. Amazon's $11 Billion Bet on AI in Georgia 💸
Amazon Web Services (AWS) is set to invest $11 billion in expanding its infrastructure in Georgia, creating at least 550 high-skilled jobs. This initiative arrives amid a booming demand for robust data centers, as tech giants like Microsoft also ramp up their investments in AI technologies, with an eye-popping $80 billion earmarked for fiscal 2025.
As AI applications continue to proliferate, the need for extensive computing resources grows, leading to a significant uptick in electricity consumption across the U.S
5. Grok's "Unhinged Mode" on the Horizon 🤬
Elon Musk's xAI is teasing a new "Unhinged Mode" for its chatbot, Grok, promising responses that are intentionally objectionable and reminiscent of an up-and-coming stand-up comic. While the mode isn’t live yet, the FAQ hints at a shift towards edginess, aligning with Musk's original vision of an unfiltered AI.
Grok already showcases a more vulgar side compared to competitors, but it has faced criticism for leaning politically left, something Musk aims to address. This development could make Grok a more daring player in the AI chatbot landscape, potentially appealing to users seeking bold dialogue.
6. 41% of Employers Plan Job Cuts Due to AI 🤯
A recent survey by the World Economic Forum reveals that 41% of employers are looking to downsize their workforce due to AI automation, marking a significant shift in the job landscape. While many companies plan to reskill their employees, the report highlights a worrying trend: jobs like graphic designers and legal secretaries are projected to decline as generative AI proves increasingly capable of handling knowledge work.
Interestingly, there’s a silver lining; demand for AI skills is soaring, with 70% of firms eager to hire talent who can develop AI tools.
Create custom GPTs that save you time!
When we first reviewed Gemini inside Google Workspace, we weren’t impressed.
But Google has been rolling out a lot of updates to its Gemini integration into Google Workspace.
So we decided to take a look at what’s new and see if the updates actually make Gemini Workspace usable.
Check out today's AI in 5.
🦾How You Can Leverage:
Your words carry weight shorties.
Think about it.
As legit every company and their mom is churning out mediocre content using LLMs, it’s your company’s own voice, data and internal decision making that helps you stand out.
Often times… those are words.
Meetings.
Speeches.
Conference calls.
The words coming out of your collective business mouths are legit paydirt.
Which leads us to an unsung hero of the Generative AI revolution — AI-powered transcription, like OpenAI’s Whisper models.
Philip Kiely gave us the lowdown on all things Whisper and AI transcription tech.
He’s the Head of Developer Relations Baseten, which provides scalable model inference and deployment for machine learning applications.
Why dive into it?
Well, your company’s data and people are becoming increasingly more valuable. And recorded conversations (and AI transcription tools!) are the best way to capture and monetize that.
Ready for the big 3 takeaways?
Let’s get it.
1 – Whisper = OpenAI’s True Open Source AI 🤖
Hold up. While everyone's drooling over ChatGPT’s next models, Philip just dropped knowledge about OpenAI's actually open source project.
Whisper, their text-to-speech transcription model.
Whisper isn't just another AI model. It's MIT licensed. Translation? You can run it anywhere. Your computer. The cloud. Your neighbor's gaming rig.
But here's the real tea: Since its 2022 launch, Whisper's error rate dropped to a mind-bending 1%.
We're talking ONE wrong word per hundred.
Perfection? Nah. But for most business use cases? Absolutely fire.
Try This:
First, download Whisper V3 and run it locally.
(As long as your computer’s got a lil GPU juice.)
No API costs. No rate limits. No nonsense.
Set up automatic transcription for your top three meeting types. Customer calls first – that's where the gold is.
Create a dead-simple tagging system: decisions, action items, customer pain points. Then watch as your team starts mining insights from conversations that used to disappear into the void.
Basemen has a good guide here to get you started in 5 minutes.
2 – Cheaper, Faster, Better = Better Business Use-Cases ✅
Philip spilled some numbers that straight-up changed the game.
Two years ago? Transcription could cost you a few bucks per hour.
Today? TWO CENTS.
But the speed gains? WILD. We're talking 1000x real-time processing. That hour-long strategy session? Transcribed before you can say "this meeting could've been an email."
The math is mathing: When you can process every word for pennies, suddenly ALL your audio becomes business intelligence.
Try This:
Start with call center gold.
There’s no reason to still only be transcribing 1% of those calls like it’s 2019, ya’ll.
Get. It. All.
Set up automatic transcription for every customer interaction.
Create specific search queries for common issues, winning solutions, and those "aha" customer moments.
Build a database of what actually works – not what people THINK works. Then use that data to train new hires, spot trends, and shut down problems before they blow up.
Cost? Less than your daily coffee habit.
3 – Making Embodied AI Possible 🧍
Philip connected some dots that'll blow your mind: Cheap, fast transcription is the secret sauce for next-gen AI interfaces.
Forget clunky voice assistants.
When transcription happens faster than human reflexes (literally – we're talking 200 milliseconds), everything changes.
Those Ray-Ban Meta glasses? Just the beginning.
The future isn't about typing into chatbots. It's about natural conversations becoming instant intelligence.
Try This:
Get ahead of this wave. Set up WhisperFlow for voice-first documentation. Train your team to capture insights via voice notes.
Build a voice-searchable knowledge base that grows with every conversation. Most importantly: Start thinking about voice interfaces for your most common tasks. Because in six months? Your competitors will be playing catch-up.
Your words are gold mines waiting to be tapped. While everyone else is copy-pasting prompts into ChatGPT, you're about to turn every conversation into actionable intelligence.
The revolution will be transcribed. Make it count.
⌚
Numbers to watch
77%
77% of large companies plan to reskill and upskill existing workers between 2025-2030 to work alongside AI.
Now This …
Let us know your thoughts!
Vote to see live results
How many AI newsletters do you read? |
Reply