What's up? You're reading Inclined AI. And we’re really starting to wonder if JJ Abrams is an advanced AI robot sent from the future to destroy franchises we love.

Anyways, here's what is on tap today:

• The state of VC investing in generative AI

• Meta launches a new AI model

• How Nvidia built their moat

• The New York Times blocks OpenAI’s bot

Be a sponsor of our newsletter.

The opportunities in generative AI feel abundant, but if you’re in venture capital, that’s far from the truth. 

VCs need huge returns to justify the money they invest in startups. In other words, a ChatGPT clone will not cut it for these investment funds.

That’s why I wanted to highlight these recent insights from a TechCrunch+ article by a VC in this space. In it, he details what areas are ripe for disruption and innovation and lays out a potential future for generative AI startups.

Let’s start with the spots in the stack that have venture-scale opportunities:

1/ Data

It feels obvious, but the companies that help researchers access more precise and detailed data will get a huge payout. Currently, no one startup dominates the market, and the problem is still nascent.

2/ Middleware

Many big companies are adopting AI models and integrating generative features into their core products. But most of that integration is customized and unpolished.

Startups that help them integrate and develop new AI applications will make revenue fast and scale to IPO levels before you know it.

3/ Fine-tuned models

Sam Altman and OpenAI are the proof here. They want to move their next models into a specialized route, and rumors indicate that even Google’s Gemini model leans this way.

But those are foundational models. The startups that use precise data and easy-to-use middleware on a model aligned to specific interest has the chance to succeed, too.

You might be wondering: what’s the logic behind picking out these parts of the stack?

It’s a combination of market size/opportunity (all these spaces are more nascent and underdeveloped) and the potential margins for effort.

For example, a new startup working on a foundational model is tempting to consider. After all, they could be the next OpenAI, right?

But, in reality, the amount of capital and time it takes for that risk to mature doesn’t align with most VC investment theses and stage emphasis.

I hope that you take this information and run with it. We know not everyone has access to this private market info, so hopefully, it gives you something you can take action on soon.

Who knows, maybe you’ll invest or build the big winner in one of these verticals.

That’d be cool.

Let me introduce you to Sonara, the true game-changer in your job search journey.

This isn't just another tool; it's the edge you've been seeking.

This marvel of AI technology is here to do the heavy lifting while you sip your coffee.

Let's talk real talk here: the job search is a numbers game. Sonara can send out over 300 job applications each month on your behalf.

That's more hooks in the water, allowing you to be more selective when those interview requests roll in.

You've got the power to ask the tough questions about benefits, company culture, and everything that matters, turning the interview process in your favor.

No more emotional rollercoaster of pouring your heart and soul into a single job application, only to be met with rejection or silence.

Sonara's got your back, making sure you're in the driver's seat.

Ready to unlock job hunting as you've never experienced?

Elevate your journey with Sonara – where your dream job is just a click away.

Meta released a new AI model called SeamlessM4T, which can translate and transcribe nearly 100 languages.

The model works for both speech and text. Some consider it a big step towards a universal translator and modern Rosetta Stone.

But SeamlessM4T is far from perfect.

The model still suffers from masculine form bias, an issue where it puts words in masculine form too often. That error can cause confusion and inaccurate transcriptions.

And then there’s the toxicity issue.

Meta’s model is impressive, but it can even work with some rarer languages that other AI tools currently struggle with. However, in some cases, the model will accidentally insert rude phrases or obscene content into translations.

You won’t run into that issue with more commonly spoken languages, but the chances aren’t zero percent.

Still, We want to give props to the AI researchers at Meta for yet another open-source model that elevates what independent developers can access. It’s a significant step forward.

🗞️ The New York Times officially blocks OpenAI’s web crawler from accessing their articles

🧠 You have the right not to have your mind read by new AI models

💰 Nvidia built a moat around AI chips, but how did they make this happen?

😳 Microsoft removes more AI-generated travel articles from their site, but don’t blame the model behind the work

🎙️ Frank Sinatra and Yo Gotti are two of the biggest influences on the next stage of generative AI

📕 ChatGPT tries to pretend that it wasn’t trained on copyrighted books now

🇨🇳 China is desperate for Nvidia chips, so desperate that they’re using worn-out ones to train their models now

🔞 WIRED publishes a controversial piece about the potential benefits of AI-generated NSFW material

Check Out Our Sunday Edition

You can dive into more AI news and topics with us every week by subscribing to our premium edition.

We’ve written about the following:

• Here Is The Work AI Will Replace Soon

• 8 Role Prompts that Still Work on ChatGPT

• 10 AI Tools You’ll Actually Use

If you’re not already subscribed, that’s okay. We’re offering a free 7-day trial so that you can read this one. That’s how excited I am to post it.

Claim the offer below to support our writing 🙂

• Rainbow - precise precipitation forecasting app (link)

• RoboResponseAI - live chat customized for your website (link)

• Kraftful - copilot for product builders (link)

• SEC Insights - revolutionizing SEC document analysis (link)

• Laxis - repurpose audio into engaging content (link)

• Hypertype - include Notion info in email responses (link)

• Ollama - easiest way to run large language models locally (link)

• Revoice - your voice, powered by AI (link)

Is This Burning Man?

I’ve never been, but clearly, AI has some idea of what goes on there. This seems accurate, right?

_{- That’s it for today. I hope you enjoyed the latest edition of inclined.ai - Davis.}