Your frustration is very real, not without reason.
But granted I feel the same angst. What's taking so long I wondered. Ironically I had to go to AI (on my phone 🫢) to find an answer. This answer sheds a bright light on why AFFU was a good fit for Affluence and Durham Black's objectives.
So hopefully it alleviates some of your frustration TJ. Nb I don't mean to insinuate AFFU/DB have a current working Generative AI build. Possible certainly, but that said their name strongly suggests they're very aware of the daunting difficulties...
~~~~~~~~~~~~~~~~~~~~~~~
Building generative AI models that work well can be challenging in various aspects, including financial and computer science difficulties. Some of the key hurdles and restrictions are:
Financial:
1. *Computational resources*: Training large generative models requires significant computational power and memory, which can be expensive.
2. *Data acquisition*: Gathering high-quality, diverse, and relevant data for training can be costly.
3. *Expertise*: Hiring experienced researchers, engineers, and developers with expertise in generative AI can be costly.
Computer Science:
1. *Mode collapse*: Generative models often suffer from mode collapse, where they produce limited variations of the same output.
2. *Training instability*: Training generative models can be unstable, leading to divergent or oscillating behavior.
3. *Evaluation metrics*: Designing appropriate evaluation metrics for generative models is challenging.
4. *Scalability*: Scaling generative models to complex tasks and large datasets can be difficult.
5. *Explainability*: Understanding how generative models work and make decisions is challenging.
6. *Adversarial attacks*: Generative models can be vulnerable to adversarial attacks, which can compromise their performance.
7. *Data quality*: Generative models are sensitive to data quality, and poor data can lead to suboptimal performance.
8. *Balancing objectives*: Balancing competing objectives, such as quality and diversity, can be challenging.
Restrictions:
1. *Data privacy*: Ensuring data privacy and adhering to regulations like GDPR, CCPA can be restrictive.
2. *Ethical considerations*: Ensuring generative AI systems are ethical, fair, and unbiased can be restrictive.
3. *Regulatory compliance*: Complying with regulations, such as those related to AI-generated content, can be restrictive.
These challenges and restrictions make building generative AI models tedious, but researchers and developers are actively working to overcome them.
Market Data 
Markets 
AI
