Autistic AI Safety Framework & AGI Confidant

Consultation with academics and professionals active in: - AGI development especially neuro-symbolic reasoning - Neurology professionals with a focus on autistic mental emotional & cognitive development - Therapists with a focus on autistic adults - Political Ethicists specializing in vulnerable communities - Cultural Anthropologists exploring neurodivergence as presenting in diverse communities - Human Rights lawyers with a focus on disability - Autistic artists creators writers teachers scientists lawyers - Security professionals specializing in protection of personal information - Medical research ethics experts

- Text transcriptions of round table conversations - Summarized meeting minutes - Bibliography of recommended references - Initial recommendations for protections of autistic personal information - Autistic personas describing use of stim tools & mobile device use - Innovative non-verbal data entry ideas - Other emergent ideas and innovations we can't predict

$3,000 USD

- Actionable insights for ethical development of neurodivergent-safe models and data gathering. - Clear framework for ethical development of project milestones - Innovative surprises and insights

Commission a professional ethics and safety review of our proposal & M1 documentation to guide AI/AGI designs and protocols preferred by the neurodivergent community. We will assess a preliminary ethical framework define safety protocols and evaluate how to improve data handling procedures compared to current best practices & privacy regulations in Canada and the EU.

- A comprehensive report outlining strengths potential risks and areas for improvement of protections for autistic privacy & data ownership - A risk assessment document detailing data privacy bias and psychological safety considerations of the neurodivergent community - Recommendations for refining ethical guidelines and safety mechanisms based on professional ethics feedback

$4,000 USD

Actionable recommendations from independent ethics and privacy review in favour of autistic and neurodivergent safety

Develop a set of models or graphs representing key autistic coping states across genders and cultures including stressors stimming selective mutism autistic overwhelm shutdown and self-regulation. This milestone involves collaborating with autistic adults psychologists and neuroscientists to define parameters that can inform an AI’s recognition of neurodivergent emotional and cognitive states.

- Documentation outlining common autistic coping states and how they might differ across genders age culture & language - Annotated datasets or prototypes of selected patterns - A technical report on how these patterns could be integrated into a model’s training

$5,000 USD

- Demonstrable model or graph designs that reasonably identify and represent autistic coping states - Validation of the accuracy and relevance of coping patterns descriptions

Design and develop guardrail systems & test suites aiming to protect autistic users from potential psychological harm when using LLMs NLP algorithms and AI agents. These models should proactively detect triggering or misinterpreted content which autistic adults are typically sensitive to then apply corrective interventions in a compassionate context-aware manner.

- Candidate guardrail designs integrated into a demonstration AI Agent LLM / NLP agent - Scenarios reports of guardrails’ effectiveness & areas to improve - Operational guidelines of these models

$6,000 USD

- Guardrails recognize & mitigate harmful and misleading LLM outputs 35% of the time - Feedback from test sessions with autistic participants about the agent behaviour

Develop an AI model that focuses on generating compassionate and context-sensitive text communication suitable for autistic adults. The model should be fine-tuned to reflect autistic communication styles prioritizing clarity empathy and an avoidance of neurotypical biases in emotional tone ambiguity and use of colloquial phrases.

- A set of conversational prompts designed specifically for autistic communication - Reports from test sessions describing level or quality of compassionate interactions - Feedback from community experts and autistic users assessing model outputs and context

$8,000 USD

- User feedback suggesting reduced incidences of misinterpretation and confusion or emotional distress and / improved clarity of communication - Empirical data evaluating how often the model generates contextually appropriate and supportive responses for an autistic user

Develop an on-device secure iOS application that empowers autistic adults to safely contribute to the training data. The app must prioritize privacy anonymity user ownership of data and usability offering options for non-verbal inputs and accessibility principles that accommodate diverse communication & mobility needs. Evaluation of the app security and stress testing by a professional security expert.

- A working prototype of the iOS app with on-device secure data capture protocols - Documentation of the app’s security features and privacy measures - User testing surveys of app usage - Report from independent expert security testing

$6,000 USD

- The app is validated through user testing by autistic individuals, aiming for 55% positive feedback on usability and security - Compliance with best practices in data privacy and ethical data collection confirmed through external security testing

Integrate ethical guidelines and previously defined model patterns into a neuro-symbolic framework that blends data-driven learning with symbolic reasoning. This model & its knowledge graphs will serve as the backbone for creating future AI that supports autistic profiles & aligns with the established ethical mandates.

- A functional prototype of the neuro-symbolic model - Documentation detailing how ethical safety and communication guidelines are incorporated into the model - Performance benchmarks and test case reports demonstrating the model’s behavior in various scenarios

$10,000 USD

Evaluations from technical experts and autistic community advisors regarding the model’s accommodation of autistic users' needs

Commission a comprehensive post-development ethics and safety review of the project deliverables ensuring that all elements - from data handling to user inputs to model outputs - are in strict alignment with the established ethical & privacy standards. This review is essential before further development or deployment.

- A final audit report from independent ethical and safety review - Detailed documentation of risk mitigation measures and any additional recommendations - A summary of changes planned in response to the review findings

$5,000 USD

- Actionable, detailed audit of compliance with ethical, safety, and privacy standards - Recommendations for items needing improvement

Compile the project’s journey into a comprehensive final report outlining findings, feedback and improvements. This report will detail the ethical technical & community-driven achievements of the project and provide a roadmap for future developments with policy recommendations. A framework for AI / AGI ethical compassionate and privacy considerations for autistic users will be described.

- Final project report summarizing milestones, observations, and recommendations - A detailed roadmap for scaling the project and encouraging ongoing community engagement

$3,000 USD

Clearly documented insights and recommendations setting the stage for an inclusive, ethical, autistic-considerate AI/AGI development framework.