How to know if an AI Crypto project is worth investing in

The integration between artificial intelligence and crypto is considered the next big thing in technology. Over the past few years we have seen AI crypto tokens reach a market cap of over $1 billion.

But despite this strong investor interest, there has not yet been a commensurate wave of consumer demand.

Ask the average AI user which program they rely on for day-to-day use, and they'll likely mention programs like ChatGPT, Brave's Leo search app, or Microsoft's Copilot.

Rarely does a user disclose whether they are using a blockchain or crypto protocol.

But will this user demand come in the future? And will blockchain AI really change the world or just the latest fundraising hype?

Cointelegraph sat down with some of the leading blockchain AI protocol developers to ask this question.

GPU demand is growing.

According to Gaurav Sharma, Chief Technology Officer of the AI ​​project IO, today's centralized cloud computing systems cannot meet the demand for graphics processing units (GPUs) that are highly demanded by AI developers, and this presents an opportunity. Decentralized blockchain projects.

Before starting the project, Sharma worked in the hotel industry, developing AI models that help predict which hotels a user will check into and what they will pay. But when Amazon asked for enough GPUs to train the model, they reportedly said they didn't have enough stock to meet his needs. And so he said.

To be honest, we went to Amazon as our first purchase. We couldn't buy it. Then we went to the cloud. We didn't get it there either, and we had to wait months at that point to get this inventory from AWS itself.

The main problem, according to Sharma, is that centralized cloud computing providers take months to set up servers at a specific location and are prohibitively expensive for the average user.

Meanwhile, there may be some GPUs sitting where the customer wants them, but because they are not owned by the supplier, they will not be delivered.

For example, if a customer goes to Amazon and requests 10,000 GPUs in Amsterdam, they will not partner with Google to provide these servers. “That's not how you do it, is it?” Sharma asked conversationally.

In his view, decentralized protocols like IO can solve this problem by creating a marketplace for GPU power.

Customers can come to the platform to find servers and vendors can offer their GPUs on the platform, allowing customers to find GPUs regardless of vendor. Given the increasing demand for GPUs as AI applications become more popular, this is the only way to effectively match buyers with sellers.

Latest: $100K Bitcoin? 9 Analysts Share Their End-of-Year BTC Price Predictions

Still, Sharma admits that some AI groups don't value blockchain AI specifically and in the broader AI industry.

Some teams claim to create the next big model with just three or five people, when in reality it takes a much bigger team than that.

Others have engineers who have worked for major companies but don't have a portfolio to show investors.

Sharma suggested that investors should be careful to investigate the team behind each project. Those who will do good work in the future have done good work in the past.

Investors should ask for the code to be open-sourced and for regular audits to let the public know how much human intervention there is in the project, he said.

Prediction markets may need AI.

According to ORA co-founder Kartin Wong, blockchain prediction markets will need to use AI in the future, which will force the integration between the two technologies.

Wong points to the rise of Polymarket as proof of this demand. As Polymarket operates on the blockchain, it “cannot have any terms to address and resolve. [the question of who won a bet]He said.

Rather, it is “often based on human judgment. But blockchain AI “can create oracles that respond to anything in the world, if that thing happens [the] Internet”

He also argued that tokenization would facilitate fundraising for AI models. ORA proposed the “initial model sacrifice”, which allowed untrained AI models to initialize tokens. The proceeds can be used to pay for the highly GPU-intensive and expensive training of the model.

According to Wong, the models launched on ORA are owned by token-holders, allowing these holders to profit from their success.

They are also open source, which creates transparency for the investing public. Wong says this solves a common problem in AI, which is that most models are proprietary so their investors can make money.

At ORA, model creators are required to abide by the licenses in open source software, which prevents developers from simply cutting and pasting code to cut creators' profits.

However, Wong admitted that there are some fake blockchain AI projects, or fake AI projects in general. Some models may claim to be generating results from AI, but they may be using humans to check the work done by the model, which makes the model far superior.

He pointed out that distinguishing between fake and real AI can sometimes be very difficult.

However, investors say the best way to assess whether a product is truly AI is to use it. He pointed to ChatOLM, a chatbot created by ORA, as an example of a product that is clearly using AI because it answers questions better than a human can.

Blockchain may allow “truly autonomous AI.”

According to Ron Chan, founder of the Blockchain AI Protocol Inference Lab, blockchain offers the only way to achieve “truly autonomous AI.” For this reason, let's look at it as a human being.

Centralized AI “depends on the goals of the enterprise,” Chan said. While this has its place in the world, decentralized AI meets a specific need.

It is “liberal – growth is driven by market demand engagement and momentum” which “creates conditions for human-centered innovation with the power to solve grand challenges.

Decentralized AI, he said, would set up systems for “verification,” or proof that a particular answer came from a specific AI model. This is an “immediate need” for the industry, he said.

Chan admits that distinguishing between human and AI projects can sometimes be difficult, if not impossible. He pointed to an example of an X user bug that claims to be under the control of an AI model.

“How can viewers be sure there isn't a human operator making decisions behind the scenes?” He asked rhetorically, pointing out that since both the AI ​​and the creator hold the account's password, it is impractical to verify who is generating the posts.

Latest: 7 Policymakers ‘Laser Focused' on $100K Bitcoin Price

The answer is to give AI sole control, verifiable independence and irrevocable delegation, Chan suggests.

AI must have “exclusive access to the account”, and third parties must be able to verify this fact.

Furthermore, once control over the account is transferred to the AI, it should be impossible for humans to regain this control. Only then can we know that anything that works is actually initiated by an AI model and not by a human working behind the scenes.

In Chan's view, this type of proven AI inference is an area that only decentralized protocols can address.

The biggest advantage may come later

Cointelegraph asked for examples of consumer-facing blockchain AI applications that users can enjoy now rather than in the future.

In response, Wong referenced the chat app OLMChat, Chan discussed the aircraft tracking AI venture and the liquid stacking app created by the Inference Labs team.

While these apps may have a small user base compared to superstar software like ChatGPT, the interviewees all had high hopes that blockchain AI will truly improve the world, even if the benefits may take some time to be fully realized by end users.