AI and Web 3.0 – a risky convergence of technologies

The advent of strong AI, or AGI, no longer seems like something coming out of a sci-fi book – it’s increasingly becoming our reality.

Naturally, this trending topic has influenced the Web 3.0 space, prompting companies and investors to focus on AI integration.

Remember the initial crypto boom when every tech company wanted to add blockchain to their tech stack, often unnecessarily?

That’s what’s happening now with AI.

Even recent major events, like ETHDenver or Token2049, which previously focused exclusively on Web 3.0, are transitioning towards a synergy of topics involving the conjunction of AI and Web 3.0.

I have no qualms about this, but being a science fiction fan has made me think a lot about the actual challenges AI brings and how we can protect ourselves from a future that looks more like the movie ‘Ex-Machina’ and less like ‘Her.’

I will share some of my thoughts on the integration of AI into Web 3.0 and how we could prevent some of the issues that could arise.

This might not happen tomorrow or next year, but it’s a future we should be prepared for.

Could ethical biases of AI models affect people’s beliefs

The first thing that comes to mind regarding ‘AI taking control over the world’ is the inherent bias of the model – who trains it and what data they use.

What happens if one trains their AI with some views or opinions that are questionable in nature?

Could this be used to influence opinions on a mass scale? Could it be used to brainwash out youth?

What if it is trained to be very subtle while doing this?

In this case, decentralization could play a dual role.

On one hand, integrating blockchain technology into AI models will make it harder for one group to take control and manipulate the system.

But at the same time, if this decentralization isn’t handled properly, it could exacerbate the challenges and easily harm the network.

Seeing the current level of discourse on X (formerly Twitter), another concern that keeps coming up is that only a few organizations will control the biases built within these AI models.

Industry giants like Meta underscore the need for open-source models while at the same time retaining control on their side.

Even OpenAI started as an open-source company and has since become anything but ‘open.’

By allowing anyone to fine-tune AI models, we can cultivate a variety of biases and perspectives.

This diversity could give people more options and prevent a few powerful groups from dominating AI development, thus promoting a more equitable technological future.

As we explore the convergence of AI and Web 3.0, it’s crucial to understand how these technologies can mutually enhance one another.

AI takes control over humans

As AI systems become more integrated into daily life, controlling and protecting personal data becomes another major concern.

In this scenario, decentralization could be quite handy, potentially giving back control to people.

However, integrating AI with blockchain comes with risks.

How can we be sure that AI is acting in the human interest?

How can we ensure that the way they have been programmed is not biased and is acting solely to support humanity?

Given these challenges, relying completely on programmed systems isn’t viable because code too can be biased.

Programming decisions can sometimes carry biases too, especially if the team is not diverse or if the data the models were trained with is skewed.

Even though technologies like AI and blockchain have a lot of benefits, we still need human oversight to ensure everything works fairly.

In a decentralized environment, human oversight could be implemented through a shared governance model.

Decisions regarding training data, model updates and ethical guidelines could be made through consensus among a diverse group of stakeholders.

This group could include developers, ethicists, user representatives and third-party auditors.

They would work together to monitor AI behavior and intervene when necessary, ensuring that AI systems align with broader human values and ethical standards.

Such a governance model not only leverages the benefits of decentralization but also mitigates the risks, ensuring that this technology enhances – rather than undermines – human welfare.

What are your main considerations regarding this AI evolution?