Can blockchain address AI’s transparency issues?

Artificial intelligence (AI) is revolutionizing industries by expanding data processing and decision-making capabilities beyond human limits. But as AI systems become more complex, they are becoming increasingly opaque, raising concerns about transparency, trust, and fairness.

The “black box” nature typical of most AI systems leads stakeholders to question the origins and reliability of AI-generated outputs. In response, technologies such as Explainable AI (XAI) have emerged to demystify AI operations, but they often fall short of fully elucidating their complexities.

As the complexities of AI continue to evolve, the need for robust mechanisms to ensure that these systems are not only effective but also trustworthy and fair increases. This is where blockchain technology comes into play, known for its key role in increasing security and transparency through decentralized record keeping.

Blockchain has the potential not only to secure financial transactions, but also to instill a layer of verifiability into AI operations that has been previously difficult to achieve. It has the potential to address some of AI’s most persistent challenges, such as data integrity and traceability of decisions, making it a critical component in the quest for transparent and trustworthy AI systems.

Chris Feng, COO of Chainbase, offered his views on the matter in an interview with crypto.news. According to Feng, while blockchain integration does not directly solve every aspect of AI transparency, it does improve several critical areas.

Can blockchain technology really increase transparency in AI systems?

Blockchain technology does not solve the fundamental problem of explainability in AI models. It is very important to distinguish between interpretability and transparency. The primary reason for the lack of explainability in AI models lies in the black-box nature of deep neural networks. Although we understand the inference process, we cannot grasp the logical significance of each parameter involved.

So how does blockchain technology increase transparency, as opposed to the interpretability improvements offered by technologies like IBM’s Explainable Artificial Intelligence (XAI)?

In the context of Explainable AI (XAI), various methods such as uncertainty statistics or analyzing the outputs and slopes of models are used to understand their functionality. However, integrating blockchain technology does not change the internal reasoning and training methods of AI models and therefore does not increase their interpretability. Nevertheless, blockchain can increase the transparency of training data, procedures, and causal inference. For example, blockchain technology enables tracking of data used for model training and incorporates community input into decision-making processes. All these data and procedures can be securely recorded on the blockchain, thus increasing the transparency of both the construction and inference processes of AI models.

Given the pervasive problem of bias in AI algorithms, how effective is blockchain in ensuring data provenance and integrity throughout the AI ​​lifecycle?

Existing blockchain methodologies have shown significant potential in securely storing and providing training data for AI models. Using distributed nodes increases privacy and security. For example, Bittensor uses a distributed training approach that distributes data across multiple nodes and applies algorithms to prevent cheating across nodes, thereby increasing the robustness of distributed AI model training. Additionally, protecting user data during inference is crucial. For example, Ritual encrypts data for inference computations before distributing it to off-chain nodes.

Are there any limitations to this approach?

One notable limitation is the oversight of model bias originating in training data. In particular, identifying biases in model predictions related to gender or race originating in training data is often neglected. Currently, neither blockchain technologies nor AI model debiasing methods effectively target and remove biases through explainability or debiasing techniques.

Do you think blockchain can increase the transparency of AI model validation and testing?

Companies like Bittensor, Ritual, and Santiment are using blockchain technology to connect on-chain smart contracts with off-chain computing capabilities. This integration enables on-chain inference, ensuring transparency between data, models, and computing power, thus increasing overall transparency throughout the process.

What consensus mechanisms do you think are best suited for blockchain networks to validate AI decisions?

I personally advocate for integrating Proof of Stake (PoS) and Proof of Authority (PoA) mechanisms. Unlike traditional distributed computing, AI training and inference processes require consistent and stable GPU resources over long periods of time. Therefore, it is imperative to verify the efficiency and reliability of these nodes. Currently, reliable computing resources are primarily hosted in data centers of various scales, as consumer-grade GPUs may not be able to adequately support AI services on the blockchain.

Looking ahead, what creative approaches or developments in blockchain technology do you foresee will be critical to overcoming current transparency challenges in AI, and how could these reshape the landscape of AI trust and accountability?

I see several challenges in current blockchain-based AI applications, such as addressing the relationship between model debiasing and data, and leveraging blockchain technology to detect and mitigate black-box attacks. I actively explore ways to encourage the community to experiment with model interpretability and increase the transparency of AI models. Moreover, I consider how blockchain can facilitate the transformation of AI into a true public good. Public goods are defined by transparency, social benefit, and serving the public interest. However, current AI technologies are often found between experimental projects and commercial products. By using a blockchain network that incentivizes and distributes value, we can accelerate the democratization, accessibility, and decentralization of AI. This approach can potentially achieve executable transparency and provide greater reliability in AI systems.