Welcome to your MetaTrends Report, a guide to help you navigate the 5 forces shaping our world.
In each issue, we look at the most important events and new opportunities. We do so through the lens of the Five Forces Framework. It’s built on recurring themes discovered through over 5,000 hours of research for more than 100+ Trends Reports.
Whether you’re a founder, investor or curious about the future, the Five Forces Framework will help you understand these key forces.
The Five Forces:
- 🦾 Individual Leverage: The increasing power of individuals to impact the world. See our reports on AI Agents, No-Code and Million-Dollar, One-Person Businesses.
- 🔓 Permissionless Innovation: The erosion of barriers to entry, allowing anyone to build and share ideas. See our reports on Paid Podcasts, The Creator Economy and Self-Publishing.
- 🔻 Specialization: The growing importance of niche expertise and targeted solutions. See our reports on Micro-SaaS, Micro-Marketplaces and Audience-First Products.
- 🏛️ Centralization: The consolidation of power and control by platforms and institutions. See our reports on Programmatic SEO, YouTube Channels and Faceless Social Media Accounts.
- 🌐 Decentralization: The shift towards distributed systems and technologies that empower individuals. See our reports on Open-Source AI, DAOs and Decentralized Finance.
MetaTrends Reports will help you:
- Stop Reacting, Start Shaping: Become an active player in shaping the future, rather than a passive observer of events.
- Gain Clarity in a World of Chaos: The Five Forces Framework gives you a sense of order in an otherwise overwhelming world.
- Turn Disruption into Advantage: See challenges as opportunities and build new solutions that others haven’t imagined.
Now let’s see the forces in action…
This free issue dives into the most important (and overlooked) events from last week. We use the MetaTrends framework to find opportunities and prepare for risks.
Hume AI: The First AI with Emotional Intelligence
The Facts:
- Hume AI is a research lab focused on building emotionally intelligent AI.
- It detects vocal expressions to sense emotions such as amusement, desire and disgust.
Opportunities:
- Build education tools. Use APIs, like Hume, to detect confusion and confidence. Use the data to personalize education. You can imagine “passive learning” by integrating with a Loom-like tool. It would share suggestions to speakers when you record an async video.
- Improve LLM interfaces. If you’ve used voice chat with ChatGPT, you’ve seen how easy it is to have long talks when you don’t have to type. What if third-party apps like TypingMind detected emotions? They could pass this as metadata to improve responses.
- Make dynamic mental health apps. If AI can persuade conspiracy believers they are wrong, it can help you break bad habits. Last week, we featured an AI therapist app that works based on transcribed words. What if underlying emotions were also detected?
- Make better games with emotionally intelligent characters. NVIDIA showed a feature with dynamic non-player characters (NPCs). This seems like a glimpse of the future with smart characters and dynamic storylines.
- Improve customer service: AI is being deployed to provide scalable customer service. Now we can enhance these systems to detect and respond based on the emotional states of humans.
- Build emotionally intelligent robots: We’re in the robotics renaissance. The robots will be emotionally intelligent. They will offer empathetic interactions with humans. Leading to new opportunities in healthcare, companionship, and elderly care.
Haters:
“The desire for things like lie detection can lead to privacy concerns.”
The idea of our innermost feelings being monitored without consent is unsettling. We lie for a reason, and it’s unsettling when an algorithm exposes us as a liar. This is a reminder that technology is a superpower. As with any superpower, individuals can use it for good or evil purposes.
“Surveillance states can use this to enhance tracking of individuals.”
We’ve seen cases of mass surveillance revealed. Nation-states often are the first to use centralizing technology. This ship may have already sailed. But this also shows the need for privacy-preserving technology. This includes end-to-end encryption and zero-knowledge proofs.
MetaTrends Analysis:
Emotion-detecting AI exemplifies 🦾 Individual Leverage and 🔻 Specialization. It lets developers build better apps without making the underlying technology. Companies like Hume are also an example of 🏛️ Centralization. All data from such developers flows through them. AI that detects emotion can also enhance mass surveillance. It does this by parsing data to find interesting elements.
Qwen 2: Alibaba’s Open-Source LLM Challenge
The Facts:
- Alibaba Cloud released Qwen 2. An open-source large language model (LLM). It beats all other open-source models.
- The tests cover natural language processing, coding, math, and multilingual abilities.
Opportunities:
- Add Qwen2 to your AI app. Developers often focus on the jobs-to-be-done. They assume that better models will come. And they’ll integrate with them. PDF.ai, Perplexity and TypingMind take this approach. If you have an AI app, add Qwen2 to the list of available models.
- Fine-tune models with Qwen2 as the base. Last week, we saw an open-source, fine-tuned model for law. Now that we have an even more powerful foundational, open-source model, you can do the same for other areas.
- Use Qwen2 for private chats. Last week, we also looked at the new space of Local LLMs. We’ve seen new use cases that are possible with better privacy. Open-source models trail centralized ones. But, the gap is closing with releases such as Qwen2. Platforms, like Ollama, make it easier to run many open-source models on your computer.
Haters:
“The model may have been trained in a biased way.”
Bias is inevitable in training. The same goes for centralized models. The only difference is that we have more access to fine-tuning open-source models.
MetaTrends Analysis:
This is the most powerful open-source model we’ve seen yet. It is an example of 🔓 Permissionless Innovation. Anyone can download and tweak the model. Open-source foundational models protect privacy. This is an example of 🌐 Decentralization. This is like other models, which can be open-source or centralized. It gives builders more power. So, it’s an example of 🦾 Individual Leverage.
Stable Audio Open: Democratizing Sound Design with Open-Source AI
The Facts:
- Stability AI launched Stable Audio Open, a free, open-source text-to-audio model for making audio clips up to 47 seconds long.
- The model can create drum beats, instrument riffs, ambient sounds, foley recordings and other sound effects.
Opportunities:
- Create ambient tracks. Some of the most popular YouTube videos are sounds that people work and relax to.
- Build sound effect libraries. This is a vertical opportunity that could be focused on podcasts, games and more. See general/horizontal libraries such as Pixabay, Mixkit and Soundsnap.
- Make custom sound effects for your games, videos and podcasts. This helps you make your brand unique without needing sound designers.
Haters:
“The model can only create short audio clips, not full songs or longer pieces.”
And this is the worst it will ever be. The rate of improvement for LLMs seems to be doubling every 5 to 14 months. Clip length will not be a problem for long.
MetaTrends Analysis:
Stable Audio Open is an open-source library. It is an example of 🔓 Permissionless Innovation and 🌐 Decentralization. It also lets non-sound designers improve products without mastering a new skill. Creators can focus on their main job. It points to 🦾 Individual Leverage and 🔻 Specialization.
In MetaTrends Reports, we highlight opportunities and events that others may have overlooked. But here are some more popular stories that you may have missed:
- Apple announces new AI features at WWDC 2024. The new features will increase 🦾 Individual Leverage. The hardware lock-in and privacy concerns tie into 🏛️ Centralization.
- Jack Dorsey recently spoke on how algorithms are programming us which means that the free speech debate should be a free will debate. He also highlights the importance of decentralized social media platforms. This is emblematic of the tug-of-war between 🏛️ Centralization and 🌐 Decentralization.
- Former OpenAI employee, Leopold Ashenbrenner, predicts Artificial General Intelligence (AGI) by 2027. He lays out the roadmap for AGI and superintelligence in an essay series called Situational Awareness. AI, with hardware and energy as chokepoints, is a centralizing technology. Leopold suggests that nation-states take more control of the technology. This is a clear case of 🏛️ Centralization.
- Former Google CEO, Eric Schmidt, is building an AI-powered military drone venture named Project Eagle. The team is testing AI-guided drones in Silicon Valley and Ukraine. Since governments have a monopoly on violence, this is an example of 🏛️ Centralization.
MetaTrends Pro #0002 has more insights:
With MetaTrends Pro you’ll learn:
- (📈 Pro) What new AI research tool changing the way we search, summarize and share knowledge?
- (📈 Pro) How to build a foundational model, like ChatGPT, from scratch in 4 hours?
- (📈 Pro) Who’s challenging OpenAI in the AI-generated video space? With a new model that’s already available to users.
- (📈 Pro) What new discovery will desalination, industrial processes and climate modeling?
What you’ll get:
- In-depth analysis of 4 more new developments and emerging opportunities
- 5 opportunities to use AI-powered research for content creation
- A comprehensive 4-hour video tutorial on reproducing a language model from scratch
- 7 applications of the new AI video model taking on OpenAI’s Sora
Upgrade to MetaTrends Pro to stay ahead of the curve and uncover hidden opportunities in the rapidly evolving world of AI and emerging technologies.
Get Weekly Reports
Join 65,000+ founders and investors
📈 Unlock Pro Reports, 1:1 Intros and Masterminds
Become a Trends Pro Member and join 1,200+ founders enjoying…
🧠 Founder Mastermind Groups • To share goals, progress and solve problems together, each group is made up of 6 members who meet for 1 hour each Monday.
📈 100+ Trends Pro Reports • To make sense of new markets, ideas and business models, check out our research reports.
💬 1:1 Founder Intros • Make new friends, share lessons and find ways to help each other. Keep life interesting by meeting new founders each week.
🧍 Daily Standups • Stay productive and accountable with daily, async standups. Unlock access to 1:1 chats, masterminds and more by building standup streaks.
💲 100k+ Startup Discounts • Get access to $100k+ in startup discounts on AWS, Twilio, Webflow, ClickUp and more.
