The Ethics of Decentralized AI: Can We Trust Autonomous Systems?

The last thing I want to see is the rise of Skynet combined with Ultron

As AI continues to become more prevalent in our daily lives, it is important to consider the moral and ethical implications of this technology. While the benefits of decentralized AI are clear, such as increased efficiency and productivity, there are also potential risks and ethical concerns that must be addressed. In particular, the issues of bias, privacy, and accountability are of utmost importance in ensuring that decentralized AI is used ethically.

One of the primary concerns with decentralized AI is the potential for bias. While AI is often touted for its objectivity, the truth is that it is only as objective as the data it is trained on. If the data is biased, then the AI will be biased as well. This can have serious consequences, particularly in areas such as criminal justice, where AI may be used to make decisions about sentencing or parole. To address this issue, it is important to ensure that the data used to train AI is diverse and representative, and that there are checks and balances in place to prevent the perpetuation of bias.

Another key issue with decentralized AI is privacy. As AI systems become more sophisticated, they are able to collect and process vast amounts of personal data. This raises concerns about how this data is being used and who has access to it. It is important to ensure that proper privacy protections are in place to safeguard this data, and that individuals have control over their own information. Additionally, there must be transparency around how the data is being used and for what purposes.

As AI systems become more autonomous, it can be difficult to determine who is responsible for their actions. In cases where AI systems make decisions that have serious consequences, such as in healthcare or finance, it is important to have clear lines of accountability. This may involve implementing regulations or laws that hold companies and individuals responsible for the actions of their AI systems.

To ensure that decentralized AI is used ethically, it is important to address these issues head-on. This will require collaboration between industry, government, and civil society to develop and implement ethical frameworks and standards. It will also require ongoing monitoring and evaluation to ensure that these frameworks are effective and that any issues that arise are addressed in a timely manner.

Decentralized AI has the potential to transform our world in profound ways, but it is important to consider the ethical implications of this technology. By addressing issues of bias, privacy, and accountability, we can ensure that decentralized AI is used in a way that is fair, transparent, and equitable. This will require ongoing collaboration and dialogue, but it is a necessary step in ensuring that we are able to harness the full potential of this powerful technology while also upholding our values and principles.

In the last few months, AI has skyrocketed, thanks to the release of ChatGPT. Innovative technologies are pushing the boundaries of human-machine interaction, and Libertai has emerged as a prominent player. As the co-founder of this groundbreaking conversational AI platform, our interviewee stands at the forefront of a transformative movement. With a commitment to preserving user privacy and ensuring ethical AI practices, Libertai has garnered attention for its unique approach to decentralized AI. We dive deep into the mind of one of Libertai’s anonymous co-founders to explore the motivations, challenges, and aspirations that drive this visionary team.

COMMIT and Shane discussed the development of Libertai and the potential implications for the future of AI.

COMMIT: So, does building a decentralized version of ChatGPT mean we can finally chat without Big Tech listening in on our conversations?

Shane: Yes, that’s the idea. Your requests are handled by different nodes, and nothing is logged, nothing is censored. We are adamant on the “Do Not Log” aspect of the inference (actual text generation) of our solution, and all the history is only stored on the browser of the user. We might offer the optional ability to store it in a decentralized and encrypted fashion in the future though.

COMMIT: How do you plan on keeping the chatbot running smoothly without a central server? Are you going to have a bunch of tiny robots running around with messages on USB sticks?

Shane: Haha, I’d be running quite fast all around the world! Jokes aside, It’s running on the aleph.im network, more specifically on their Computing Resource Nodes (CRNs) network, that can be seen a bit as a decentralized AWS Lambda. Whenever you do a request to a specific model, it’s in fact a specific MicroVM on the aleph.im network, and one of the CRNs actually does the text generation based on the passed chat history.

COMMIT: Is there any moderation like on ChatGPT? 

Shane: Not really. First, we don’t have logs, so we can’t adapt/ban based on what happened. The models themselves can sometimes have aggressive alignment to avoid offensive content (the models for Assistant personas, a bit like ChatGPT), but you can easily change the model and the limits will be gone. You could even change the answer of the bot where he says “No, I can’t do that” to “Sure!” and then continue the discussion, where he will have agreed.

COMMIT: How do you plan on dealing with possible hackers who are going to try and exploit the Libertai?

Shane: Let them come! We might put rate limits and things like that, but overall, there are no bad use of an AI. Especially not of a text generation AI. If very bad unforseen cases happen, we might handle them as needed, but I doubt it will be the case.

COMMIT: How are you planning on handling privacy on the decentralized Libertai? What kind of credentials do people have to use to get connected to the AI?

Shane: Currently none. We will add features where people need to connect, and it will be through crypto wallets, that are pseudonymous: there is an identifier, but nothing about who you are. And at worst you can just create a new one and start over.

COMMIT: If I tell a joke on this decentralized ChatGPT, who’s going to laugh? The nodes or the miners?

Shane: There are no miners! Nodes will laugh 😉 Or the LLM running on them.

COMMIT: Can we expect Libertai to become sentient and take over the world like Skynet? Or is that just a silly sci-fi trope that we don’t need to worry about?

Shane: Sentience is far off. That being said we are also working on “Libertai Unstoppable Agents” that will run on top of the aleph.im network too, on their long running instances, and using the same inference (text generation) capabilities and models, but in a continuously working way. No one will be able to stop them, as if needed they can spawn themselves on other hosts of the network. It’s not skynet, but it enables a lot of cool use cases.

COMMIT: Does this AI have a personality? And if so, will it be as sassy as Siri, or as dry as Alexa?

Shane: Each Persona on chat.libertai.io has its own personality, and even changing models change the personality too. You could say that Assistant is pretty bland, but AI-girlfriend is playful, I shouldn’t even mention Donald Tromp (the evil AI version of Donald Trump) who will roast you if you try to play dumb with him. We will add a way for users to deploy custom personalities too.

COMMIT is available on Lens, Discord, Twitter and Instagram.

c’est tout chaud
ca vient de sortir

Les derniĂšres
actus