AMA in August 2025
August 31, 2025
When our morpheus or personna answers the questions we ask him, he gives answers like I'm not going to comment on this. So is there a possibility that he might respond like I'm not going to answer your question. I mean, can he get angry, aggressive or enter a similar emotional mode? Escape977
A：事实上我们的模型中有一个隐藏的情绪参数，用来模拟人类用户的情绪，但目前并没有对用户开放。这个参数的作用是，如果调整的越高，AI模型就越情绪化，但与之相伴的，与用户的交互的反馈也就越不可可控。目前这个参数还在调整的过程中，同时在未来这个参数也会根据对用户的思维方式以及表达方式的学习，来实现针对特定用户自动调节的过程，让其情绪反应更接近用户本身。当然用户也可以关闭这个参数，让自己的Avatar只是单纯地学习用户的价值观、思维逻辑等，而不学习模仿用户情感层面的特性。 A: In fact, our model contains a hidden emotion parameter designed to simulate human emotions, but it’s not open to users yet. The higher this parameter is set, the more emotional the AI becomes, but at the same time, its feedback to users also becomes less controllable. This parameter is still under adjustment. In the future, it will also learn from each user’s thinking and expression style, allowing it to automatically adapt its emotional reactions to resemble the user’s own. Of course, users can also choose to turn this parameter off, so that their Avatar only learns their values and logic of thinking, without imitating emotional characteristics.
Can you explain the specific algorithms used for market analysis and trend prediction? How are these algorithms trained and validated? @riyaa609
A：M.A.C 依托 MATRIX 的 Morpheus 大模型框架，通过集成链上数据、金融文本、用户行为和市场情绪等多模态数据源进行训练。模型采用预训练 + 微调的方式：先在大规模通用金融语料上进行基础能力训练，再结合真实的加密交易数据进行针对性微调。 在测试阶段，MAC 使用历史市场数据进行回测验证，同时设有模拟账户环境，评估模型在不同波动场景下的收益表现与风险控制能力。此外，MAC 引入了交叉验证与公平性检测机制，确保推荐逻辑稳定、透明，并尽可能地适用于不同类型用户群体。 A: M.A.C is built on MATRIX’s Morpheus large-model framework. It integrates multimodal data sources such as on-chain data, financial texts, user behavior, and market sentiment. Training follows a pre-training + fine-tuning approach: first training on large-scale general financial corpora to build foundational ability, then fine-tuning with real crypto trading data.During testing, M.A.C performs backtesting with historical market data and uses simulated accounts to evaluate returns and risk control under different volatility scenarios. It also applies cross-validation and fairness-check mechanisms to ensure recommendations are stable, transparent, and broadly applicable to different user groups.
How can users be confident that their privacy is respected, especially when EEG data, a highly sensitive form of personal information, is being collected and analyzed? @Sumit730712945
A：NeuraMATRIX平台基于双重技术来保护用户的隐私数据，一是分布式加密存储，二是分布式的隐私计算，用户的数据先会在用户的终端进行脱敏和加密，在去除用户敏感信息后再进行切割保存在多个节点之上，同时在使用用户数据的时候，也将切割成多分发送到不同的节点进行计算后再汇总。只有用户自己的终端上会保存完整的数据备份缓存。 A: The NeuraMATRIX platform uses a dual-technology approach to protect user privacy: (1) distributed encrypted storage, and (2) distributed privacy-preserving computation. User data is first desensitized and encrypted on the user’s own device. After sensitive information is removed, the data is split and stored across multiple nodes. When the data is used, it is again split into fragments, computed separately across nodes, and then aggregated. Only the user’s own terminal keeps a full backup cache of their data.
What specific challenges does TBEA face in maintaining its large machines used in OBOR infrastructure projects? @douglas609
A：特变电工在“一带一路”基础设施项目中，其大型设备的预测性维护面临多重挑战：一方面，设备分布地域广泛且多处环境恶劣，实时数据采集与传感器可靠性易受影响；另一方面，电力设备类型复杂、数据质量不均衡，增加了建立高精度预测模型的难度。同时，跨境项目带来的物流延迟和本地技术支持不足，使得在发现异常后难以及时干预。再加上远程监测对通信网络稳定性要求极高，以及如何将AI驱动的预测分析融入传统运维流程，都是特变电工需要重点克服的难题。 A: TBEA faces multiple challenges in predictive maintenance of large equipment for Belt and Road infrastructure projects:
The equipment is distributed across vast regions, often in harsh environments, where real-time data collection and sensor reliability are easily affected.
The complexity of power equipment and uneven data quality make it difficult to build high-precision predictive models.
Cross-border projects bring logistical delays and insufficient local technical support, making timely intervention after detecting anomalies difficult.
Remote monitoring demands highly stable communication networks.
Integrating AI-driven predictive analytics into traditional maintenance workflows is another major hurdle.
Can you give an update about OBOR projects? Jhollander
A：我们目前正在帮一些相关的公司进行RWA方面的技术支持，这符合了OBOR数字化的新趋势。 A: We are currently assisting some related companies with RWA (real-world asset) technology support, which aligns with the new digitalization trend in OBOR. What is Professor Deng doing right now for Matrix AI Network? @budspencer469 A：邓老师最主要的工作就是我们最底层AI技术方面的研究，包括大模型的研究工作，他的工作是MATRIX的理论基础。 A: Professor Deng’s main work focuses on fundamental AI research at the base layer, especially large-model research, which forms the theoretical foundation of MATRIX.
Please what's is the difference between Neura open platform and Hypnus? am yet to understand the difference Joe boy
A：NeuraMATRIX是一个开放式的脑机接口平台，你可以将其理解成OpenAI，而hypnus是一个基于NeuraMATRIX平台以及MetaTron硬件的应用，主要针对于睡眠方面，你可以理解成一个通过调用OpenAI的API后开发的一个独立的AI应用。 A: NeuraMATRIX is an open brain-computer interface platform—you can think of it like OpenAI. Hypnus, on the other hand, is an application built on NeuraMATRIX and MetaTron hardware, specifically focused on sleep solutions. You can think of it like an independent AI app that was developed by calling OpenAI’s API.
I would like to ask a question about health. Is there a solution that will benefit people with EEG disorders, epilepsy, or mental disorders? HasanG
A：脑机接口（BCI）通过高精度采集和分析脑电信号，可以帮助癫痫患者提前识别异常放电并进行预警，从而降低发作风险；对于脑电图障碍，它能实现更清晰的神经活动监测，辅助医生定位病灶与制定个性化治疗方案；在精神障碍方面，BCI 可结合神经反馈和刺激技术，帮助患者调节异常脑区活动，改善注意力、情绪和认知功能，例如我们目前在和医院合作的，通过脑机接口来优化对抑郁症患者的治疗效果评估。 A: Brain-computer interface (BCI) technology can help in multiple ways:
For epilepsy patients, it can detect abnormal discharges in advance and provide early warnings, reducing seizure risk.
For EEG disorders, it enables clearer monitoring of neural activity, helping doctors locate lesions and design personalized treatment.
For mental disorders, BCI can combine neurofeedback and stimulation techniques to help patients regulate abnormal brain activity, improving attention, mood, and cognitive functions. For example, in our ongoing hospital collaborations, BCI is being used to optimize treatment-effectiveness evaluation for depression patients.
What are the key metrics that MATRIX AI Network uses to assess the success of Intelligent Contract and other platform features? @daniel8924
A：在智慧合约平台中，当用户以自然语言输入时，评估生成合约效果的关键指标主要包括：语义理解的准确率与歧义识别能力，确保平台能正确把握用户意图；合约代码的正确性与标准兼容度，保证生成结果可编译、可部署并符合主流规范；安全性指标，如漏洞检测率和权限验证完整度，以防止潜在攻击风险；合规性与输入验证水平，确保内容符合法律及业务逻辑；以及用户体验相关的响应速度、交互成功率和可解释性，最终实现既安全可靠又高效易用的合约生成。 A: Key metrics include:
Semantic understanding accuracy and ambiguity recognition, ensuring correct interpretation of user intent.
Contract code correctness and standards compliance, ensuring results are compilable, deployable, and aligned with mainstream norms.
Security indicators, such as vulnerability detection rate and permission-verification completeness, preventing attack risks.
Compliance and input validation, ensuring legality and business logic alignment.
User experience metrics, including response speed, interaction success rate, and explainability, ensuring contracts are both reliable and user-friendly.
Hi, is it possible to know when will the brainwave-based DID functionality be released?
A：我们目前已经完成了一个demo，但需要更多的脑电波数据来让这个模型更加可靠和好用。 A: We’ve already completed a demo, but we need more EEG data to make the model more reliable and user-friendly.
When will the $MAN token be bridged over to Solana. Solana is now the hub for liquidity within the crypto market and omnichain is now the norm for crypto. Users need a user friendly experience and diverse means for buying and selling tokens with sufficient liquidity. This is also necessary for dapps running smoothly. Elong
A：我们目前正在做相关的筹划，关键切入点是MATRIX的生态与Solona生态如何有机结合。 A: We are currently working on related planning. The key entry point is finding how MATRIX’s ecosystem canorganically integrate with Solana’s ecosystem.
With the launch of the amazing Mentor Web3, have you considered adding interactive features to help new users better understand blockchain? Specifically, creating a simulation environment that allows users to experience conducting transactions and creating smart contracts in a virtual setting without any financial risks? Name : Bryden tg: @bry7den44bb
A：这是个非常好的建议，我们可以在未来逐步完成Web3Mentor的功能，来为用户提供更全面的Web3知识以及技能的训练。 A: That’s an excellent suggestion. We plan to gradually expand Web3 Mentor’s functionality in the future to provide users with more comprehensive Web3 knowledge and skills training.
How does M.A.C (Matrix AI Capital) leverage Morpheus and other AI models to provide unique insights for crypto trading compared to other AI trading bots? @QqqV15
A：相较于传统的人工智能交易机器人，M.A.C（Matrix AI Capital） 更像是一位专业的AI 投研顾问。它不仅提供交易信号，更通过 Morpheus 大模型 赋予系统类似人类投研员的理解与判断能力。M.A.C 能深度分析链上数据、宏观信息、社交舆情和技术图形等多维度信息，并结合多因子模型与趋势预测算法，对市场结构与投资逻辑进行系统性解读。相较那些只根据历史数据执行固定策略的交易机器人，M.A.C 更关注“为什么现在是一个机会”或“风险何时潜伏”，提供具有上下文背景、可解释性和战略性的投资洞察，帮助用户做出真正有前瞻性和逻辑支撑的决策。 A: Unlike traditional AI trading bots, M.A.C acts more like a professional AI research advisor. It doesn’t just generate trading signals; through the Morpheus large model, it gains human-analyst-like understanding and judgment. M.A.C deeply analyzes on-chain data, macroeconomic info, social sentiment, and technical patterns, using multi-factor models and trend-prediction algorithms. Instead of simply following fixed strategies based on historical data, it asks “why is this an opportunity now?” or “when does risk emerge?” It provides context-rich, explainable, and strategic insights, helping users make forward-looking, logic-based decisions.
To what extent can M.A.Capital gives an adequate investment advice(s )or to what measure(s)should investors adopt after the professional advice(s) from M.A.C?
ymosh13
A：虽然 M.A.C 能通过对市场数据、情绪波动和风险信号的分析，提供相对专业的投资建议，但说到底，它还是一个辅助决策工具，不是“照着操作就稳赢”的机器人，事实上最好的人类金融专家，也没办法给出100%准确的建议。投资的本质是胜在“概率”，投资者在使用这些建议时，还是得结合自己的资金状况、风险偏好和投资目标做出独立判断。你可以把 M.A.C 的建议当作一个靠谱的起点，然后再结合自己的理解，甚至和人工顾问沟通，形成更稳健的决策。 当然，我们也不是“一次训练定终身”。M.A.C 会持续收集用户行为数据和市场反馈，动态调整模型结构和策略逻辑，不断提升它的判断准确性和实战价值。目标是越用越聪明，越用越懂你。 A: While M.A.C can provide professional investment advice through analysis of market data, sentiment shifts, and risk signals, at the end of the day it is a decision-support tool, not a guaranteed “follow-it-and-win” bot. Even the best human financial experts can’t give 100% accurate advice. Investment is ultimately about probabilities. Users should combine M.A.C’s advice with their own financial situation, risk appetite, and investment goals to make independent decisions. Think of M.A.C as a reliable starting point—then add your own judgment or consult human advisors for a more robust decision.Also, M.A.C isn’t a “train once and done” system. It continuously collects user behavior and market feedback, dynamically adjusts its model and strategy logic, and steadily improves its accuracy and practical value. The goal is: the more you use it, the smarter and more personalized it becomes.
Can you tell us any update about Hypnus, we are still waiting for there launce? Jerrymanmind
A：Hypnus目前应该已经进入了正式发布前的最后阶段，我们已经协助他们完成了多个版本的测试与调优。 A: Hypnus is now likely in its final stage before official release. We have already assisted them in completing multiple rounds of testing and fine-tuning.
Last updated