- The Rundown AI
- Posts
- Control your dreams using AI
Welcome, AI enthusiasts.
Think controlling your dreams sounds like Science Fiction? A new AI-powered device called Morpheus-1 may change your mind.
The halo-like headband claims to help induce and control lucid dreaming — and if its upcoming trials succeed, our sleep may never be the same. Let’s explore…
In today’s AI rundown:
New device aims to induce lucid dreams on-demand
OpenAI rolls out new model upgrades, cost cuts
How to access Copilot's GPTs & plugins on mobile
Pika releases RPG for improved prompt comprehension
6 new AI tools & 4 new AI jobs
More AI & tech news
Read time: 4 minutes
LATEST DEVELOPMENTS
PROPHETIC
Image source: Prophetic
The Rundown: Tech startup Prophetic just introduced Morpheus-1, described as the world's first ‘multimodal generative ultrasonic transformer’ designed to explore human consciousness through controlling lucid dreams.
The details:
Morpheus-1 combines machine learning, EEG data, and ultrasound waves to detect REM sleep and trigger dream states that users can control.
A halo-like headband detects brain waves, with ultrasound beams then stimulating specific brain areas tied to conscious dreaming.
Onboard batteries, Bluetooth, and WiFi allow all-night operation and data syncing, with a companion app tracking sleep phases and collecting dream data to refine the AI models.
Prophetic said the beta program for Morpheus-1 will begin this spring.
Why it matters: Ready to control your dreams? With so much still left to understand about the brain and sleep, combining AI with neuroscience could be a major unlock for demystifying and expanding human consciousness.
TOGETHER WITH BRAVE
The Rundown: Building the next great AI? Take it to the next level with Brave Search API:
Power your AI with real-time queries from an index of 20 billion pages (with millions more added daily)
Use Brave's unique, user-driven data, not just web crawlers, to train your LLM and chatbots with real-time data
Enjoy an intuitive data structure at a fraction of Bing's cost
Your AI is nothing without a connection to the web. Click here to get started with 2,000 free queries/month.
OPENAI
Image source: Midjourney
The Rundown: OpenAI just announced several new upgrades to its GPT models, including reduced ‘laziness’ and lowered API costs alongside other technical improvements.
The details:
GPT-3.5 Turbo gets unspecified improvements, a 50% input cost drop, and a 25% output price cut for API usage.
GPT-4 Turbo gets an upgrade to reduce ‘laziness’, addressing complaints of the model’s quality degrading.
The top tier 4T model with Vision will also be launching in ‘general availability’ in the coming months.
Developers gain more control over API keys, with the text moderation API also getting improvements to catch harmful content better.
Why it matters: Reduced API costs add up for developers building off OpenAI tech, as well as help keep them in the ecosystem as competition swells. The holidays are apparently over for GPT-4, getting a 2024 attitude adjustment after recent reports of the AI underperforming.
AI TRAINING
Image source: Microsoft Copilot
The Rundown: Microsoft Copilot just got a new mobile update, rolling out the ability for users to access both GPTs and plugins for free in-app on the go.
Thank you to Paul Couvert for sharing this tutorial.
Step-by-step:
Open the Copilot app — or download it if you haven’t already via the App Store or Google Play Store.
Access GPTs in the menu bar on the left side of the app — the four currently available are Designer, Vacation Planner, Cooking Assistant, and Fitness Trainer.
Enable plugins by clicking the three-dot options icon at the top right and selecting ‘Plugins’ — you can choose up to three at a time to activate.
AI RESEARCH
Image source: Pika
The Rundown: Pika researchers just developed a new approach called RPG (Recaptioning, Planning, Generating), which uses powerful multimodal AI systems to enhance text-to-image models, improving generations from complex text prompts.
The details:
RPG leverages chain-of-thought reasoning and multimodal understanding, breaking down prompts into simpler sub-prompts and planning complementary regions for each subset.
The images are then generated region-by-region, guided by the subprompts.
The approach significantly outperforms leading diffusion models on metrics like text-image alignment and multi-category object composition.
Why it matters: While text-to-image models have come a long way over the last year, they still struggle with capturing every detail of a complex prompt. Pika’s latest breakthrough brings an impressive level of control — giving creators even more power over their outputs.
NEW TOOLS & JOBS
🎥 guidde- Magically create stunning video documentation with AI (link)*
🎤 Tech Interviewer- A fully-simulated AI interviewer that acts like an actual human (link)
📸 Keep it shot- Instantly rename and find messy screenshots with AI (link)
💬 Dubbing Studio by ElevenLabs- Precise control over translating videos for global audiences (link)
⚠️ Norm AI- Eliminate compliance risks for your business at scale with AI (link)
🗣️ izTalk- Overcome language barriers with instant AI translation (link)
*Sponsored listing
QUICK HITS
Google Cloud and Hugging Face announced a new partnership to advance open-source AI, with HF users now able to leverage Google infrastructure to build models.
Brave integrated open-source Mixtral 8x7B as the default LLM on both free and paid versions of its AI browser assistant Leo.
The Federal Trade Commission is probing the investments of tech giants like Microsoft, Amazon, and Google into OpenAI and Anthropic, investigating how the deals impact the AI competitive landscape.
PayPal revealed plans for new AI products, including tools for merchants to better reach new customers and a Fastlane feature for ramping up checkout speeds.
FuboTV launched "Instant Headlines," a feature that uses AI to generate 10-word summaries of news segments to help viewers decide what to watch in real-time.
University of Amsterdam researchers developed RoboChem, an autonomous AI robot for chemical synthesis that outperforms human chemists in speed and accuracy.
AI-generated graphic images of Taylor Swift went viral across X, outraging fans and sparking broader discussions around restricting deepfake content.
THAT’S A WRAP
SPONSOR US
Get your product in front of over 450k+ AI enthusiasts
Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners around the world. Get in touch today.
FEEDBACK
How would you rate today's newsletter?Vote below to help us improve the newsletter for you. |
If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.
Join the conversation