Tag: ai

Artificial Intelligence, Blockchains, cryptography, Decentralized, Patents, Science, Uncategorized, Yogi Nelson

Access to technology is a human right, not a copyright

Yogi Nelson

Welcome to the BlockchainAIForum where your technologies questions are answered. Today, part three of the Unbound Science project is on the menu. In this final series, I cover the project in greater detail, including: challenges facing science, maintaining organizational governance and financial independence, platform economics, and more!

At Unbound Science we believe scientific research has never been free to truly explore the unknown and we have made it our mission to unbind science. Throughout the course of human society, science has been subjected to control for reasons that include ideological, political, religious, and of course economic. The Unbound Science platform encourages imagination and fringe experimentation. However, some knowledge could be catastrophic in the wrong hands. Potentially dangerous research proposals, including but not limited to military weapons, experimentation with black holes, nuclear energy, and DNA modification, therefore will not be permitted. 

Challenges Facing Science: Unbound Science proposes to address the six most pressing challenges facing independent science research:

Corruption: Until now, science has been dominated by agenda-based research. Facts have been created in controlled experiments, as opposed to funding open experimentation to discover the actual truth. Unbound Science has come up with a way to disintermediate the donor’s agenda.

Security: Undoubtedly, scientific research that destabilizes the status quo can jeopardize the security of scientists. Scientists who disrupt society through their inventions could face security threats from business/government incumbents who fear losing oligopoly market power, or ideological fanatics who are enemies of progress.

Preservation: Stop for a second and consider all the science humanity has lost due to destruction, be it natural or man-made. Lost science is a crime against humanity. Unbound Science is committed to do its part to end the tragedy of lost science research.

Finance: Financial independence is nearly impossible for the vast majority of scientists. Often scientists spend inordinate time soliciting grant applications and thereafter submitting monitoring reports to the handful of granting institutions.

Subjugation: Unbound Science envisions an end to the current system where new technologies allow one society to take great advantages over all others, and remove their freedoms and even their own ability to advance as a society. It has taken generations and even centuries for subjugated peoples to even get back to zero, let alone be able to compete on the global stage. By empowering the entire workforce of 8 billion people on the planet with the same access to tools and capabilities, human progress can accelerate forward like never seen before.

Distribution: One marketplace to share plans and rights across the world, in a single click! Scientists can generate recurring revenue from different contracts, including consulting, licensing, and leasing their IP. They can track all their clients and royalty payments from one place, instead of having to worry about cheap knock-offs and lawsuits hampering the true representation of their work.

Solutions to the Challenges Facing Science

Complex problems require comprehensive solutions. Hence, Unbound Science proposes an open-source scientific research platform, driven by and for the community. Our summarized solutions to the sxi challenges are listed below:

Addressing Corruption: We surveyed which sectors could provide the largest amount ofdisruption and benefit to mankind, and have not yet been adequately addressed by traditional science, Open-Source Science, or DeSci yet. Therefore, we’ve selected four areas to start, with more potentially being added later. At platform launch, benefactors may elect to contribute to one of four pools:

  • Energy
  • Food Technology (air, water, soil)
  • Pollution (focus on clean-up)
  • Quantum Research

Instead of a benefactor donating directly to a project listed on a proposal board, the funds are donated to a pool. These funding pools are controlled through votes, with different weights given to donors, scientists, partnership science foundations, and the Open-Source Science community. These donation pools are powered by cryptocurrency, and legally owned by a tax-optimized entity for issuing tax write-offs to donors (where applicable).

Addressing Security: Scientists may opt to stay anonymous or pseudonymous, building an on-chain reputation and be judged solely by the merits of their work, both for their physical protection and for the sake of fairness. If you can do the work, your formal educational background (or employment) does not matter. Of course, a scientist with an already prestigious track record may choose to display their full personal information, for the sake of obtaining funding and asserting a greater influence.

Addressing Preservation: To participate, scientists agree to upload all research data, notes, blueprints, and final instructions for replication to blockchain(s). Scientists can choose a variety of mechanisms for when to begin sharing this data, such as keeping it private until their work is complete, or their life is presumed to have ended.

Addressing Finance: The platform features a “YouTube-like” interface, with a game-show-atmosphere experience for scientists to pitch concepts to a rotating panel of judges, who host the funding competition in seasons, each with several phases. Scientists no longer have to fill out an enormous application and wait six months for a few thousand dollars. They can simply submit a 5—8 minute video describing their idea. Winners in phase one are awarded their requested budget to develop a Proof of Concept. Winners in Phase 2 are given their requested budget to create a final prototype ready for mass production. Winners in Phase 3 are awarded their requested budget to create a mass production factory and further innovation laboratory.

Addressing Subjugation: In return for funding, scientists must declare their inventions open-source, and share all of their experiment data, processes, and all other documentation for anyone to use and copy without restrictions. Winning scientists will earn revenue from the creation of “originals”, consulting contracts, and leftover general prize money. One of the other benefits of releasing open-source technology is that scientists do not also have to become international business owners in order to achieve success. By sharing the technology, they do not need to open their own factory and deal with all the costs and headaches of starting a global manufacturing enterprise. Their vision becomes reality a lot faster, cheaper, and there are still plenty of profits to be made for the scientist.

Addressing Distribution: A marketplace is available for sharing buttoned-up packages of building plans and licensing agreements for worldwide production and distribution by third-parties. Open-source licensing for individuals, leasing of IP in exchange for royalties from corporations and governments, as well as consulting contracts to ensure the production is done at the highest of quality. Scientists can track all their business relationships through one place.

Maintaining Organizational Independence:  Voting Caps and Weights

To prevent any particular group from dominating the process, Unbound Science has decided to put limits on the number of tokens that can be used to vote, per project. For instance, there will be a cap of 100,000 tokens for general users, 150,000 for admins, 200,000 for scientists, 500,000 benefactors.

Unbound Science seeks to allow everyone to have a voice in what gets brought into the world. However, some voices should be amplified given their dedication, commitment, actions and participation. Hence, user actions can achieve milestones to unlock new features, higher voting caps and voting weights. The actual milestones will be finalized after the beta launch, and based on community suggestions. Below is a hypothetical example:

The benefactor limit is 500,000 tokens at a time with which they can use to vote on a single proposal. If a benefactor reaches a milestone of $1m USD donated, their voting cap raises beyond 500,000, to 1,000,000. Actual milestones will be finalized after the beta launch, and based on community suggestions.

Further examples of specific actions that might unlock higher caps: user votes in every season within a calendar year, scientists get nominated to game show, scientists become game show finalists, scientists win the game show, scientist data passes peer review, users hit pre-determined milestones for watching videos (e.g. 1,000, 5,000, or 10,000 hours etc.) and Admins successfully filter out 100+ pieces of spam and/or fake data.  Weights (counted as a multiplier of tokens used in their vote)

  • Top 30 Most Active Benefactor Wallets: 2x
  • Scientists who have contributed research: 1.5x
  • Scientists who have been game show finalists: 2x
  • Scientists who have won the game show: 3x
  • New scientists this season: 1.4x
  • New community members this season: 1.2x
  • Scientists with more than 1 year of participation: 1.1x
  • Community members with longevity: 1.0x
  • Admin votes: 1.0x
  • Wallets with transactions from to banned/suspended/muted users: 0.5x for the amount sent to them.

Maintaining Financial Independence:  Handling Donations

Benefactors send USDC, USDT, DAI, ETH, WBTC, MATIC, BNB, AAVE, LINK, COMP, UNI, MKR, or SNX to the inbound vault. The price of assets are calculated against the benefactor’s local currency, and a 503c donation document is generated and emailed to the donor. Non-stable assets are deposited into Aave, Compound, and Venus vaults, where they will be used as collateral to borrow stablecoins if needed to replenish the pool’s spending, otherwise they will be left to earn interest and appreciate over time. Stable assets will transfer to cold storage via MPC vaults. This will be a manual process, handled by platform management, on behalf of the pool owners (consortium of non-profit organizations).

Empowering Scientists

The current science funding system is sub-optimal and characterized by long applications, needless forms, and busy-body gatekeepers that take months or years to come to a decision on whether or not to sign off on financial support. The Unbound Science way gathers funding from the public, then lets members of the public and scientific community work together to vet and allot financial support in a more streamlined, transparent, and easy to participate manner.

Like a TV show, funding is handled in seasons. Each season has a show-host and a panel of judges, selected from the community, by the community.

  • The first phase of a season qualifies proposals from scientists or requests from the community.
  • The second phase of a season announces approved proposals from Phase 1. Preliminary funding is requested by the scientist and given approval or adjustment by the judges handling this season. This funding is intended to give a scientist support for a Proof-of-Concept or Minimum Viable Product.
  • The third phase votes on which scientists will be funded to finish solving the problem. Not every team will be ready at the same time, and may delay their participation into future seasons whenever they are ready, without further qualification.

Approved funds will be sent in USDC or USDT to scientists via newly generated vaults (i.e., Cardano), separated by team, and per project. This helps minimize security risks and also makes accounting an easier process than if all funds lived in the same vault. Stablecoins can be off-ramped into fiat, through Ethereum bridges like Jagon and Wanchain. Funding in vaults is unlocked during various epochs, to ensure lapses in development and abandoned projects are not rewarded. More details on epochs below.

Progress Reporting

Approved and funded proposals have a page created where they can be tracked publicly, with regular updates and achieved milestones required to qualify for the release of the team’s next epoch of funding. Data must live on a non-censorable decentralized third-party platform; Notion or Asana links are considered invalid. Updates are received in PDF format which is indexed by IPFS, Arweave, and Spacemesh for immutability. Physical storage (not just the metadata) is suggested to live inside web3-native storage enclaves like Internet Computer or Iagon. Upon an update being accepted by the community, funding for the next epoch is unlocked and made available.

Research essentials must be uploaded and stored on-chain as open-source with the full intent of helping future generations of scientists. This includes materials lists, notes, diagrams, blueprints, schematics, code, graphics, trials data, metadata, and anything else of importance (excluding names of participants and sources of materials).

User Accounts

Every user signs up with a general community account – everyone starts in the same place.

There are 4 types of user accounts:

  • General Community
  • Scientists
  • Donors
  • Admins

General Community users can earn platform tokens for:

  • Watching videos
  • Liking videos
  • Commenting
  • Hitting milestones in discussion forums and on-site time
  • Consistent voting during live shows

Scientists can earn platform tokens for:

  • Uploading videos
  • Uploading documentation
  • Participating in peer review discussions
  • Being selected to participate in the game show (as contender or judge)
  • Advancing to the 2nd or 3rd round in the game show
  • Successfully voting out scams/fake content
  • General Community actions

Benefactors can earn platform tokens for:

  • Making contributions to donation pools
  • General Community actions

Admins can earn platform tokens for:

  • Removing spam/fake content
  • Being voted in as a game show judge
  • More General Community actions to be determined at a later date

User accounts are private by default. Nevertheless, users may elect to make certain elements of their profile and user data available for the public to see. For instance, users may link existing web3 wallets via WalletConnect, rather than making a whole new identity. Users may self-identify their account type, and fill in profile information if they wish for this information to be shared with the public.

Benefactors

The ideal benefactors are passive donors looking for tax incentives, and the scientifically curious. Traditional science funding sources, e.g. governments, foundations, and private institutions are welcome and encouraged to participate.

Unbound Science recognizes that traditional funders may opt out lacking control over decision authority. As an entry point to facilitate a transition, Unbound Science may encourage traditional funders to consider the Visionary NFT acquisition option. This option may be the preferred route, for passive revenue sharing and aligning incentives. The platform also encourages “retail users” looking for tax write-offs and seeking to apply their money towards making the world better, faster.

Upon confirmation of the transaction, the donor’s wallet will receive an NFT certificate that functions as a receipt and states the amount of the donation with a thank you message.

A yearly leaderboard will track the top five donors and they will be issued a special NFT, which is soul-bound to the donor’s wallet. These special NFTs are likely to attract airdrops and invitations from other projects who wish to do business with scientific donors. Furthermore, these NFTs will be minted on Ethereum main net, as it is most prestigious and likely to attract quality airdrops. The soul-bound aspect means they cannot be sold or moved. They will be minted/destroyed as the leaderboard updates. It is a temporary digital badge/trophy, as opposed to the tax-break receipts, or scientific data dynamic NFT containers, which are immutable and forever.

Benefactors will be informed and must comprehend a critical fact: Unbound Science works as a pool of liquidity. Once donated, benefactor funds will be pooled together into one commingled account. In other words, the design of Unbound Science is for benefactors to provide funding to a liquidity pool, as opposed to a common experience on similar donation websites, where each individual funding their own accounts and then choosing where to grant those funds. Unbound Science is doing things differently. Each of the four starting subject matters (Energy, Pollution, Air/Water/Soil, Quantum) have their own liquidity pool from which the community votes how much to give in grants and to whom.

Example:

Whale A donates $100, and receives 100,000 number of voting tokens. With his/her tokens Whale A decides to vote for Proposal A, using all 100,000 tokens. Community donations are pitiful this season, and raise only an additional $100. Community receives 100,000 tokens split among participating members. Community votes to reject Proposal A, but approves Proposal B to receive 100% of the season’s pool. Community participants had a higher voting weight so the vote swung in favor of the community. Also voting caps limited Whale A’s total vote to 1% of the entire vote. Whale A’s entire donated funds have been allocated to Proposal B. Whale A’s gets refunded and his/her new balance sits at 999,000 for future seasons.

Scientists

Current research is funded by institutions and universities which then retain all the rights and intellectual property. Just consider when was the last time we heard about a person or small team of people inventing something? Almost never. It’s always this corporation or that university. Unbound Science returns credit and control back to the individuals and groups who put in the work. Everything is undergoing a digital decentralized revolution; science should not be left behind.

Scientists will sign up on the platform with their personal information and credentials via third-party plugins. Once verified, the information will be encrypted and stored on-chain. Scientists may selectively divulge information to the public, or remain anonymous, only showing a pseudonym and degree status. Scientists may check or uncheck whether or not they are willing to collaborate with other scientists who have similar proposals or ongoing research. Scientists who are willing to collaborate are given a special badge on their profile, and an extra 10% voting cap over those who are unwilling. Universities who are recognized leaders in research and research fundings, who have led the way previously, will be invited to participate in this new way of doing things, even though this method of conducting fundraising may be laughed away at first, as was Bitcoin in finance, or cloud computing in tech.

Technology Distribution

When the research is complete, scientists will upload their compiled data to a marketplace where it can easily be licensed by global manufacturers or governments, or may be used freely if the scientist elects. The scientist can keep track of their consulting contracts, royalty payments, licensing fees, and open-source contracts all in one place.

Under The Hood

Unbound Science is a web-based application that rides on top of several different web3 technologies. The scientific research itself requires rigid and publicly supported blockchain with years of decentralization and proven security, while identity, data storage, marketplaces, and IP management are served up from a redundant array of options.

Currently involved in the development stack: Secret Network, Flux, Optimism, Fileverse, Story Protocol, Internet Computer, Arweave, Evoke Marketplace, and more. Essentially, anything that can be connected via Secret Network’s “Secretpath” between Ethereum, Solana, and Cosmos can be easily integrated.

Emerging Market

The Unbound Science team conducted comprehensive research on market conditions and possible competitors. We discovered two critical facts and those were the guiding lights as we formulated our project. First, there is a unmet need in the marketplace for this service. The evidence is the existence of four other market participants in this space. Second, as described below, existing market participants are not truly comparable to Unbound Science on matters that are materially important. For instance, only Unbound Science proposes decentralization, security, community, and sensible tokenomics as core principles. Hence, Unbound Science is the only known project capable of meeting the market demand. The other market participants are as follows:

  • ​Recerca​ – fundraising tool for research. They do many things very well, including winning 2nd prize at the Hedera X Filecoin Grant Program. The shortcomings Recera suffers is insufficient decentralization by design. Moreover, the Recera project does not feature tax incentives and they failed to solve the headaches of needlessly lengthy, dull and monotonous funding applications. Council still acts as gatekeepers to donation.
  • ​Experiment.com​ – fundraising tool for research. Donors can browse research proposals and causes, and donate in accordance with their concerns. This project resembles a kickstarter marketplace design. The project does not adequately solve centralization issues, nor application issues, nor is it built on web3 technology that can operate independently. Furthermore, there are no associated tax incentives.
  • ​inTheory Research Markets​ – funding tool for research and reputation. The platform allows scientists to create patents and pieces of research as NFTs thus facilitating research financial transactions. Last, the platform allows licensing or purchasing of intellectual property (IP).
  • ​Molecule​ – Similar to Experiment.com, but focused only on BioMed research. Raised a $13M seed.

Platform Tokenomics. Tokens are created for voting and earning achievements on the platform. These are not intended to be used for investment, speculation, expectation of profit, trade, or ownership of Unbound Science. The community may use their accumulated pile of tokens to vote for their favorite science proposals. Gamified tasks will be a fluid evolution to continuously keep the community engaged, but may represent milestone achievements through hours of video watched, comments made, finding hidden easter eggs around the site, etc.

Total Supply: 1,000,000,000 platform tokens

Platform Token Emissions: To be determined.

“It’ll take a lot more than words and guns
A whole lot more than riches and muscle
The hands of the many must join as one
And together we’ll cross the river.”
– Puscifer, The Humbling River

Follow us on X ​@unboundscience​

Until next time,

Yogi Nelson

Uncategorized

Unbound Science: Revolutionizing Decentralized Research

Yogi Nelson

Welcome to the Blockchain & AI Forum, where your technology questions are answered.  Last week’s article introduced the decentralized science concept.  This week, and next, an example of a decentralized science project–Unbound Science–is presented.  Where to start?  Why not with the Unbound Science mission statement, then a brief explanation of why the name Unbound Science, and concluding with the six problems Unbound Science was designed to solve.  Next week, the mechanics of Unbound Science will be detailed

What is Unbound Science.  Unbound Science is the next upgrade to science itself.  The mission of Unbound Science is to fund, preserve, and distribute the critical information required to replicate new technologies, and spread them around the world faster than ever before.  “Access to technology is a human right, not a copyright”, is the tag line and it captures the essence of Unbound Science. 

What’s In a Name.  According to its founder, Aaron Malone, he choose the name Unbound Science to reflect a problem-solving ethos.  What problems are in the cross-hairs of Unbound Science? The six problems listed below, says Unbound Science Founder Malone.

Security.  Undoubtedly, scientific research that destabilizes the status quo can jeopardize the security of scientists.  Scientists who disrupt society through their inventions could face security threats from business/government incumbents who fear losing oligopoly market power, or ideological fanatics who oppose progress.  The lack of security has far too often translated into a constraint on the human imagination.  Unbound Science wants to shatter restrictions on science and scientists.  Unbound Science’s goal is to lift the gag order regardless of the origin and allow scientists to imagine, innovate, and invent, without fear or reprisal.  Let’s explore the remaining five reasons Unbound Science; they are all equally compelling.

Corruption.  Until now, science has been dominated by agenda-based research.  Facts have been created in controlled experiments, as opposed to funding open experimentation to discover the actual truth. 

Preservation.  Stop for a moment to consider all the science humanity has lost due to destruction, be it natural or man induced.  Lost science is a crime against humanity.  Unbound Science is committed to do its part to end the tragedy of lost science research.

Finance.  Financial independence is nearly impossible for the vast majority of scientists.  Often scientists spend inordinate time completing grant applications and thereafter submitting exhaustive monitoring reports to the handful of granting institutions.

Subjugation.  Unbound Science envisions an end to the current system where new technologies allow one society to take great advantage over all others, and remove their freedoms and even their own ability to advance as a society.  It has taken generations and centuries for subjugated peoples to reach the starting point, let alone be able to compete on the global stage.  By empowering the entire global workforce with equal access to tools and capabilities, human progress can accelerate at an unprecedent pace.

Distribution.  Unbound Science is a cyber marketplace where scientists can share plans and rights across the world with a single mouse click.  With Unbound Science, scientists can partner with firms that specialize in production and distribution thus allowing them to focus on science rather than factory operations.  Moreover, scientists can generate new and recurring revenue across various platforms, including consulting, licensing, and leasing their intellectual property. 

Next week the mechanics of Unbound Science platform will be revealed, including its tokens, governance, operations, etc.  For now, I say goodbye but first a proverb from Gabon, where they say: “the tree that forgets its roots will fall when the storm comes.”

Until next time,

Yogi Nelson

AI Agents, AI Tools, Artificial Intelligence, Blockchains, Science, Uncategorized

Understanding Decentralized Science: A New Era for Research Funding

Yogi Nelson

Welcome to the Blockchain & AI Forum, where your technology questions are answered.  Is there an alternative to the existing science research funding model? The answer is yes–there is an alternative, it’s called Decentralized Science (DeSci) and the best example is called Unbound Science.  We begin with what is science research funding, who funds science, and conclude with a very brief mention of Unbound Science.  An exhaustive examination of Unbound Science will be presented in parts two-and-three of this three-part series over the course of the next three weeks.

What is science research funding? There is no precise definition of science research funding that is universally accepted.  However, using a general consensus definition, I think we can say science research funding:  

is the financial support provided to researchers and institutions to conduct studies, experiments, and investigations that push the boundaries of human knowledge. This crucial resource enables scientists to explore new ideas, develop cutting-edge technologies, and tackle some of the world’s most pressing challenges. It typically covers expenses such as personnel salaries, laboratory equipment, data collection, publication costs, and travel for conferences

Who Funds Science Research.  Please draw your attention to the critical phase in the definition, “… financial support provided…”.  The statement begs the question, “provided” by whom?  The answer to that question follows.

According to the National Science Foundation, science research funding by category and percentage of the total science budget is: 

  • Private Sector, i.e., Corporate Research and Development                           50%
  • Federal Government Agencies                                                                       35%
  • Academic institutions                                                                                     11%
  • Private foundations                                                                                            3%

Clearly the top three funding sources dominate the landscape.  If you add the four funding sources the sum equals 99%.  Where is the missing one percent?   You will find it in something called emerging decentralized science platforms.  What are decentralized platforms within a science research funding content?  I’ll explain that next and start by naming this phenomenon—it’s called decentralized science or DeSci.     

What is Decentralized Science (DeSci). DeSci is an emerging movement and technological framework that uses blockchain, cryptographic tools, and decentralized governance to improve how scientific research is funded, conducted, validated, shared, and rewarded. DeSci aims to address long-standing inefficiencies and inequalities in traditional science by enabling open, transparent, and community-governed research ecosystems.  DeSci operates in the web-3 ecosystem; thus, separating it from existing models.  Web-3 ecosystems are characterized by decentralized technologies, i.e., smart contracts, tokens, and governance via decentralized autonomous organizations (DAOs), not top-down hierarchies.  In DeSci, the emphasis is fund, govern, and distribute scientific research in a transparent, open-access, and community-driven manner. In other words, when operating as intended, DeSci is antithesis to the status quo and disrupts the big donor science research funding model. 

How Does DeSci Work.  Under ideal circumstances DeSci is characterized by the following six traits:

  1. Funding Research via Blockchain.  Scientists can raise funds directly from across the globe using crypto-currencies, NFTs, or project-specific tokens.
  2. Decentralized Governance.  Funding decisions can be made through DAOs and/or community members.
  3. Tokenized Incentives.  Contributors are rewarded with tokens for participation, publication, peer-review, data sharing, and many other activities.
  4. Open Access and Data Transparency.  Research outputs are stored on decentralized storage platforms, making the research permanently accessible.
  5. Reputation and Credentialing.  With DeSci verifiable credential, on-chain peer review, and reputation scores are all possible and that helps assess the credibility of researchers without relying solely on traditional academic gatekeepers.
  6. Interoperable and Modular.  Generally speaking, DeSci platforms are composable.  Composable platforms allow interoperability across funding tools, DAOs, publishing platforms, and decentralized identity systems.

What is Unbound Science.  Unbound Science is a prime example of a DeSci platform.  Unbound Science is designed to check every box on the checklist above when it reaches full maturity. Next week a deep dive into Unbound Science.

Until next time,

Yogi Nelson

AI Agents, AI Tools, Blockchains, cryptography, Uncategorized, Yogi Nelson

Building Effective AI Agents: A Complete Guide

Yogi Nelson

Welcome to the Blockchain & AI Forum, where your technology questions are answered.  Today’s question: are there practical tips for building an artificial intelligence (AI) agent?  Answer—yes.  Today I’ll explain, starting with what is an AI agent.

What is an AI Agent.  AI agents are poised to become ubiquitous.  Says who?  All the IT giants, e.g. Amazon, Microsoft, Google, etc., that’s who. Let’s start by defining AI agents.  According to IBM, AI agents:

a system or program that is capable of autonomously performing tasks on behalf of a user or another system by designing its workflow and utilizing available tools.  AI agents can encompass a wide range of functionalities beyond natural language processing including decision-making, problem-solving, interacting with external environments and executing actions. AI agents can be deployed in various applications to solve complex tasks, including software design, IT automation, code-generation tools and conversational assistants. AI agents use advanced large language models (LLMs) to comprehend and respond to user inputs.

Knowing When to Build an AI Agent. Open AI, the company behind ChatGPT, says building AI agents requires rethinking how existing systems make decisions and handle complexity.  Therefore, if you find yourself dealing with the situation below, you have an ideal circumstance for introducing AI agents, says Open AI:

Click to access a-practical-guide-to-building-agents.pdf

  1. Complex Decision Making.    Workflows involving nuanced judgment, exceptions, or context-sensitive decisions, e.g. refund approval in customer service workflows.
  2. Difficult to Maintain Rules.  Systems that have become unwieldly due to extensive and intricate rule sets, making updates costly or error-prone, e.g. performing vendor security reviews.
  3. Heavy Reliance on Unstructured Data.  Scenarios that involve natural language, extracting meaning from documents, or interacting with users conversationally, e.g. processing a home insurance claim.

AI Agent Design Foundations.  Fundamentally, AI agents consists of three core components: 1) model; 2) tools; and 3) instructions.  Let’s briefly define each core component.  Model refers to the large language model powering the agent’s reasoning and decision making. Tools means the external functions or APIs the agent can use to take action.  And instructions are the explicit guidelines and guardrails defining how the agent behaves. Next we examine principles for selecting a model.

Principles for Choosing a Model.  Not all AI models are identical.  Models vary by capabilities.  And, of course, not every task requires the best model.  Ideally build your AI agent with the most capable model for the task, factoring in trade-offs related to task complexity, latency, and cost.  With those considerations in mind, use the following methodology:

  1. Set up evaluations to establish a performance baseline.
  2. Focus on meeting your accuracy target with the best model available.
  3. Optimize for cost and latency by replacing larger models with smaller ones where possible.  If you want an Open AI model, visit this link:  https://platform.openai.com/docs/guides/model-selection

What are AI Tools.  AI tools extend the capabilities of AI agents by using APIs from underlying applications or systems.  Tools should have a standardized definition, enabling flexible, many-to-many relationships between tools and agents.  AI agents need three types of tools:

  1. Data.  Data enables AI agents to retrieve context and information necessary for executing workflow.
  2. Action.  Action tools enable agents to interact with systems to take actions, i.e., adding new information, updating records, or sending messages.
  3. Orchestration.  This is where it gets a bit science fiction.  Orchestration allows AI agents themselves to serve as tools for one or more AI agents!  When to use multiple agents?  When the single agent model fails to follow complicated instructions or consistently selects incorrect tools.

Best Practices for AI Agent Instructions.  Without high quality instructions, your AI agent will be suboptimal.  Below are a few suggestions for creating high quality instructions:

  1. Use existing documents. 
  2. Prompt the AI Agent to break down the tasks into smaller more manageable steps.
  3. Define clear actions.  In other words, make sure every step corresponds to a specific action.
  4. Capture edge cases.  Not everything fits in a box and sometimes information is missing.  Hence, instructions should anticipate common variations and include instructions on how to handle the non-routine with conditional steps.

Guardrails.  What are guardrails?  Essentially, guardrails are layered defense mechanism.  When properly designed, guardrails help manage risk.  Consider three factors when constructing guardrails:  1) privacy and safety; 2) add new guardrails based on real world experience; 3) optimize the balance between security and user experience.  With that in mind below are possible AI agent guardrails:

  1. Relevance Classifier.  This ensures the AI Agent stays within the intended scope by flagging off-topic queries.
  2. Safety Classifier.  These detect unsafe inputs that attempt to exploit system vulnerabilities.
  3. PII Filter.  PLL filters prevent unnecessary exposure of personally identifiable information.
  4. Moderation.  Moderation guardrails flag harmful or inappropriate inputs.
  5. Tool Safeguards.  With tools safeguard you can assess the risk of each tool available to the AI Agent.
  6. Rules-Based Protections.  The idea behind rules-based protection is to use simple deterministic measures to prevent known threats.  
  7. Output Validation.  Ensure responses align with brand values via prompt engineering and content checks.

Until next time,

Yogi Nelson and his AI Agent

AI Agents, Artificial Intelligence, Blockchains, cryptography, Uncategorized, Yogi Nelson

Understanding AI Safety: Global Initiatives Explored

Yogi Nelson

Welcome to the Blockchain & AI Forum, where your technology questions are answered!  Today’s question: is anyone working on artificial intelligence (AI) safety initiatives?  Let’s begin at the top—what is AI safety?

What is AI Safety.  According to IBM, AI safety involves ensuring AI systems are used responsibly, developed with human values, and that potential risks are identified and mitigated. It also includes safeguarding confidentiality and integrity of AI models, addressing bias, data security, and vulnerability to external threats.  AI safety should minimize potential harm to human and the environment, as well as prevent misuse and other unintended consequences.  

AI Safety Institutes Across the Global.  While researching this topic, I discovered the existence of numerous AI safety initiatives and institutes on every continent.  Some are government dominated, others industry led, and you can also find AI safety institutions within the non-profit/research eco-system.  Today, I’ll highlight a few AI safety initiatives and institutes from every continent.  Where to start?  Why not in alphabetic order with the African continent.

AI Safety Initiatives in Africa.  For better or worse, there is no unified AI safety initiative coming out of Africa, as a continent.  However, that doesn’t mean individual African nations are asleep.  To the contrary, several African nations have active AI safety programs.  For example, Nigeria has a draft national AI strategy in circulation that includes AI safety.  https://paradigmhq.org/wp-content/uploads/2021/11/Towards-A-Rights-Respecting-Artificial-Intelligence-Policy-for-Nigeria.pdf   Or for instance, in Egypt the National Council for Artificial Intelligence, a government think tank, produced the following report on AI that includes AI safety.

Click to access AIstrategy%20English%2016-1-2025-1.pdf

Numerous other African nations have strong initiatives but for now we move on to Asia.

AI Safety Initiatives in Asia.  It’s unrealistic to expect that Asia would be limited to only one umbrella organization dedicated to AI safety.  Asia is to diverse and enormous of the continent for that.  Nevertheless, there are regional blocks of nations working in collaboration to ensure safe AI.  A prime example is Artificial Intelligence Safety Asia, AISA.  AISA is a global non-profit dedicated to the international governance of safe AI by Asians, starting in Southeast Asia.  Their goal is to build local expertise and foster collaboration in order to produce practical AI safety policies.  AISA accomplishes the mission by training civil servants, forging partnerships, and “incubating” junior researchers.  https://www.aisafety.asia/

To no surprise, Singapore has a major AI safety institute; it’s known as Artificial Intelligence Singapore Institute (AISI).  AISI has several AI safety initiatives underway, including an extensive multi-year effort with the Digital Trust Centre Singapore (DTC) and the Infocomm Media Development Authority of Singapore (IMDA) for the purpose of advancing AI safety in Singapore.  https://sgaisi.sg/

AI Safety Initiatives in Europe.  Europe has a plethora of AI safety initiatives, and one of those is the European Network for AI Safety, ENAIS is its acronym.  ENAIS is a think tank of researchers from across Europe united in the effort to advance safe AI use.      https://www.enais.co/

AI Safety Initiatives in North America.  The U.S. government has an AI safety institute; it’s called the U.S. AI Safety Institute (US AISI).  The mission is to identify, measure, and mitigate the risks of advanced AI systems, says USAISI.  The idea is to harness the enormous potential of AI technology by developing the testing, evaluations, and guidelines required to accelerate trustworthy AI innovation–with a keen focus on helping to prevent misuse by those who seek to undermine public safety and national security.  https://www.nist.gov/aisi

AI Safety in South America.  According to a recent study by the Brookings Institute, a think tank in Washington, D.C., regional cooperation in the field of AI safety is emerging in Latin America.  Brookings notes that Brazil, Peru, and Costa Rica, are all using existing AI safety models from Europe, North America, and Asia as well as their own ideas to create AI safety standards in their respective nations.  https://www.brookings.edu/articles/regional-cooperation-crucial-for-ai-safety-and-governance-in-latin-america/

Problem solving begins with recognition that a problem exist, or may soon appear.  Its encouraging to see a concerted effort across the global working on AI safety.  Time to say goodbye with the World Economic Forum’s motto: “entrepreneurship in the global public interest”.  Lol. 

Until next time,

Yogi Nelson