Solari: How We Use AI
Become a member: Subscribe
NTS Headlines
Headlines by Category
Headlines by Year
- Money & Markets
- Weekly Solari Reports
- Cognitive Liberty
- Solari Builders
- Ask Catherine
- News Trends & Stories
- Equity Overview
- War For Bankocracy
- Digital Money, Digital Control
- State Leader Briefings
- Food
- Food for the Soul
- Future Science
- Health
- Metanoia
- Solutions
- Spiritual Science
- Wellness
- Building Wealth
- Via Europa
Solari’s Building Wealth materials are organized to inspire and support your personal strategic and financial planning.
Building Wealth
This 6-pillar curriculum teaches the basic literacy we need to be personally and financially successful and to do so in a manner in which together we evolve a culture that supports the emergence of an advanced human civilization.
Missing Money
Articles and video discussions of the $21 Trillion dollars missing from the U.S. government
Connect with the wider network of like-minded Solari subscribers using our in-house social media platform, Solari Connect.
No posts
“Some people worry that artificial intelligence will make us feel inferior, but then, anybody in his right mind should have an inferiority complex every time he looks at a flower.”
~ Computer scientist Alan Kay
Solari: How We Use AI
I. SOLARI REPORT
A. Goal #1
A primary goal in our use of AI is to understand AI, how it is used, and its impact on all aspects of our subscribers, their families, our business, and the world.
B. Goal #2
A second goal is to apply and manage its application in our business and lives, and determine risks and how to manage them.
Our Approach
AI is useful but not reliable—it is equivalent to a junior employee who requires adult supervision and can occasionally go off the rails, and even do serious damage.
Anything AI-generated or AI-supported is reviewed and approved by a team member before it goes anywhere.
Subscriber and client data are not to be touched or accessed by AI.
No Solari team member is required to use AI if they prefer not to use it.
We consider all digital technology inherently problematic and dangerous, as its development and financing were designed to enhance and support central control. The question before us is whether digital technology can evolve to something useful and/or whether it is possible to function in modern society without it. We agree with many of the warnings related to AI—but we believe they apply to all digital technology, not just AI. AI potentially accelerates the risk of central control in numerous ways.
AI companies in the United States are offering their products on a heavily subsidized basis. We appreciate the danger of sharing data with and becoming dependent on AI offerings that may not be economically available in the future.
We intend to keep analog options alive and well in our lives and business and encourage others to do the same.
AIs Used
- Chat GPT
- Claude
- Whisper
- Descript
- Midjourney (sometimes Freepik)
- Gemini
- Dart’s built-in AI—project management, being phased out in favor of Plane (self-hosted)
C. Functionality: How We Use AI
1. Search Engine
- Case law and statutes
- Domestic and global timelines
- Current events
- Research topics
- Fact-checking
- Statistics
- Travel information
Information we rely on or include in publications is fact-checked and reviewed by a team member or trusted contractor.
2. Legislation and Rulemaking Review and Drafting
3. Report Analysis and Summary
- For internal use
4. Contract Review and Drafting
- Not for confidential or sensitive information
5. Business Letter Review and Drafting
- Not for confidential or sensitive information
6. Brainstorming
- Fleshing out arguments
- Generating lists and outlines on a particular issue
- Developing titles
- Compliance and administrative checklists
- Marketing materials
7. Spell Check and Editing
8. Creating Featured Images and Original Artwork
9. Audio and Video Editing
- Generating video-related transcripts
10. Software Development
- Generating code for website functions and posting
- For CRM automation
- Problem-solving
11. Creating Graphs, Charts, and Slides
12. Project Management and Tracking
13. Note-Taking and Organization
D. How We Don’t Use AI
We do not use AI to create our content. Specifically, AI is not used for:
- Writing articles and commentaries
- Content written by guest authors: People who write for us are requested not to use AI for writing
- Generating video and audio content
E. Commercial Relationships
Based on early due diligence, we obtained two annual subscriptions for Claudi.ai—one for the Solari Report and one for Solari Investment Screens. Based on subsequent disclosures, we intend to move away from Claude and are researching alternative options.
We have asked for and reviewed proposals from one company to use open source AIs to provide an improved search and account function for subscribers at Solari.com. We intend to continue to look for a way to do so.
We anticipate that the continued drops in price and energy usage will open up greater ability to use open source AI.
II. SOLARI INVESTMENT SCREENS
Solari Investment Screens uses AI in the following ways:
- Identification of company stocks for acquisition and related reports for team member review
- Monitoring existing universe of approved stocks and related reports for team member review
- Refinement of the screening framework
- Building and maintaining a dashboard to manage both acquisition and maintenance
- Research on industries, countries, and assets
We should note that no large-cap AI or data center companies, or other companies associated with building the digital control grid, have been approved for inclusion in the Solari World Screen.
III. COMMENTS
We welcome comments on our current use of AI. Subscribers can post in the comments section or send comments to intel@solari.com
Log in or subscribe to the Solari Report to enjoy full access to exclusive articles and features.
Already a subscriber?
Thanks for this.
I’ll begin by mentioning that I’ve had advice that Claude AI is the goto source for writing programming code.
I currently run Ollama using one of several models held locally on one of my machines. While running deepseek-coder-v2 Ollama reports that of the 146GB loaded, 21% is run on the GPU and the remainder processed by the CPU.
I use this very selectively. Sometimes, I can only describe the task a similar to pulling teeth. I have to find other ways to get it to do simple tasks.
As an example, I asked an earlier model to tell me how many letter ‘r’s were in the word ‘raspberry’. I was never able to get it to return ‘three’ as an answer.
With another model, after much questioning, I was able to get it to admit that some vaccines contain a live virus.
I get the feeling that the bigger the AI model, the more options there are for things to go wrong.
I’ve explored the Huggingface site a bit. The only work I ran there was to produce images from text. Part of the task was to get an image of Plato’s Cave from a few lines of text. I’d suppose, objectively, the models worked well enough.
I’m left with the impression that AI isn’t that smart.