- AI In The Middle
- Posts
- Gen AI Brings Back the Dead
Gen AI Brings Back the Dead
Jesus shreds and likes fresh powder too 🏂
Now that I got your attention (thank you Midjourney!), welcome to our 10th issue. We have awesome stuff to tease your appetite.
What do we have for you today?
Gen AI for DeathTech — how generative AI is being used to keep relatives “alive” and explore history. Generative FOMO — how the international community is getting in on, and in some cases ahead of, generative AI. Ebooks for Everyone — Apple’s AI narration is bringing down the barriers of entry to the audiobook market for writers of all kinds.
Talking To Dead People
DeathTech has a vertical gaining steam thanks to AI. Though not a new concept (Replika was founded in 2017 for instance), the fascination of keeping relatives “alive” and talking to historical figures is gaining more momentum in the generative AI space. In a time where genealogy, legacy, and ancestry proliferates, new products allowing us to chat, interact with, and still feel connected to the departed are popping up to help with the cause.
Humans crave connection and it’s essential for our wellbeing. For many, our closest relationships are with those no longer with us. Gen AI is being used to preserve precious memories, build new ones (as if they never physically left), and educate and explore the minds of elders. Some Gen AI is advancing products so much that we’re even able to ask them if they have questions for us.
Who would you talk to? What do you think, too creepy or something you’d like? I’m imagining corporate Lunch & Learns with Steve Jobs. Evening chats with Maya Angelou or my Grandma. Casual DMs with Tony Stark as if I'm Peter Parker (fictional characters count too, no?).
Generative FOMO
Much of the western world was going loco with ChatGPT last month (no shame, myself included). Some communities were feeling the FOMO though. As a work-around and benefit from the hype, people have found ways to bypass the restriction, they’re creating their own versions (like ChatGPT Pro, an unofficial ChatGPT/GPT3 app doing $200k in revenue as I type and is #1 in France), and entire countries like China forming their own generative AI technologies.
AI is a field that is developing lightning fast. More and more of the global community will get their hands and minds on generative AI. It’ll be interesting to see what wins, what dies, and what ideas come about. It’ll be a winner-take-all situation if done right.
At the same time, regulations and legislation will be in full-force this year. We already know of the EU AI Laws being reviewed right now, China already has banned deepfakes, and the U.S. is figuring this AI stuff out too. To my fellow Americans don’t worry or get too excited. You and I both know how long it takes to get anything done around here.
"Once upon a time in a land far, far away"
The barriers to become an audiobook author have just been lowered. Apple released its AI narration on Friday and its quality is impressive. The tool allows smaller publishers and indie writers to join the growing audio marketplace and convert their written pieces into audio easily.
With digital voices that sound shockingly human, text-to-audio narration is having its Stable Diffusion moment. Some are already saying it’s much better than Polly by AWS. As speech synthesis and NLP models progress, we can imagine more literature being made available in audio and speech (beyond English too).
Bet you $5k right now, someone in SF or NYC is already having an LLM like ChatGPT write a book for them and plotting on getting it AI-narrated with Apple Books as you read this.
New developments to spam your #random Slack channel 💬
⚖️ DoNotPay is willing to pay you (or your attorney) +$1,000,000 (or more!) to have its Gen AI powered lawyer fight your case. That's how serious founder Joshua Browder is.
TikTok is capitalizing on generative Avatars
PromptLayer, the first-ever prompt engineering platform
Futurepedia: an AI directory updated daily
OpenAI's supposedly valued at $29B now having some people very excited and others rolling their eyes
Things to learn when you need a raise
Muse, Google’s text-to-image model, achieving state of the art performance and outdoing diffusion models (research paper here)
Evaluating LLMs ethical and moral behaviors and outputs — surprisingly smaller and non-RLHF models performed better (Twitter thread; research paper).
Still asking, “what’s reinforcement learning with human feedback or RLHF?” Here you go.
Want access to the Generative AI investors database? Share and subscribe to AI In the Middle (below) to receive it! ⬇️
Before you go
Apparently you can guilt trip ChatGPT, quasi-Karen style.
That's it for today folks! Thanks for reading and sharing, our AI community is growing more and more every single week.
— Matthew J. Sánchez (@matthewjsanchez)