The dark side of Web3: how decentralization fuels AI bias

Artificial intelligence (AI) has rapidly changed the way we live and work. Still, the challenge of AI data bias has come to the fore. As we move towards a Web3 future, it is only natural that we will see new innovative products, solutions and services that use both Web3 and AI together. And while some commentators argue that decentralized technologies may be the answer to data bias, that’s no less true.

The size of the Web3 market is still relatively small and difficult to quantify as the Web3 ecosystem is still in its early stages of development and the exact definition of Web3 is still evolving. While the market size in 2021 was estimated to be close $2 billionseveral analysts and research firms have reported an expected compound annual growth rate (CAGR) of approximately 45%, which combined with the rapid growth of Web3 solutions and consumer adoption puts the Web3 market on a course that will be about worth it $80 billion by 2030.

Although it is growing rapidly, the current state of the industry combined with other factors in the technology industry is why bias in AI data is on the wrong track.

The link between bias, quality and volume

AI systems rely on large amounts of high-quality data to train their algorithms. OpenAI’s GPT-3, including the ChatGPT model, is trained on a huge amount of of high-quality data. The exact amount of data used for training has not been disclosed by OpenAI, but is estimated to be hundreds of billions of words or more.

That data was filtered and pre-processed to ensure it was of high quality and relevant to the language generation task. OpenAI used advanced machine learning (ML) techniques, such as transformers, to train the model on this large dataset, enabling it to learn patterns and relationships between words and phrases and generate high-quality text.

The quality of AI training data has a significant impact on the performance of an ML model, and the size of the dataset can also be a critical factor in determining the model’s ability to generalize to new data and tasks. But it is also true that both quality and volume have a significant impact on data bias.

Unique risk of bias

Bias in AI is an important issue as it can lead to unfair, discriminatory and harmful outcomes in areas such as employment, credit, housing and criminal justice.

In 2018, Amazon was forced to scrap an AI recruiting tool that showed prejudice against women. The tool was trained on resumes submitted to Amazon over a 10-year period, including predominantly male applicants, which caused the AI ​​to downgrade resumes containing words like “female” and “woman.”

And in 2019, researchers found that it is commercially available AI algorithm used to predict patient outcomes was biased against black patients. The algorithm was trained on predominantly white patient data, which resulted in a higher false positive rate for black patients.

The decentralized nature of Web3 solutions coupled with AI poses a unique risk of creating bias. The quality and availability of data in this environment can be challenging, making it difficult to accurately train AI algorithms, not only because of the lack of Web3 solutions in use, but also because of the population that is able to to use them.

We can draw a parallel with the genomic data collected by companies like 23andMe, which are biased against poor and marginalized communities. The cost, availability, and targeted marketing of DNA testing services such as 23andMe limits access to these services for individuals from low-income communities or those living in regions where the service is not active, which are typically poorer, less developed countries.

As a result, the data collected by these companies may not accurately reflect the genomic diversity of the wider population, leading to potential biases in genetic research and the development of health care and medicine.

And that brings us to another reason why Web3 is increasing the preference for AI data.

Industry bias and focus on ethics

The lack of diversity in the Web3 startup industry is a big problem. From 2022, women will hold 26.7% of technical jobs. Of those, 56% are women of color. Executive positions in tech have an even lower representation of women.

In Web3, that imbalance is exacerbated. Less than 5% of Web3 startups have a female founder. This lack of diversity means AI data bias is likely to be unconsciously ignored as a problem by male and white founders.

To meet these challenges, the Web3 industry must prioritize diversity and inclusion in both data sources and teams. In addition, the industry needs to change the narrative of why diversity, equality and inclusion are necessary.

From a finance and scalability perspective, products and services designed from different perspectives work for billions of customers rather than millions, making startups with diverse teams more likely to achieve high returns and global scale. The Web3 industry should also focus on data quality and accuracy, to ensure that the data used to train AI algorithms is free of bias.

Can Web3 provide the answer to AI data bias?

One solution to these challenges is the development of decentralized data markets that enable secure, transparent exchange of data between individuals and organizations. This can help reduce the risk of data bias as it allows for a wider range of data to be used when training AI algorithms. In addition, blockchain technology can be used to ensure data transparency and accuracy so that algorithms are not biased.

But in the end, we will face the great challenge of finding broad data sources over many years until Web3 solutions are used by a mainstream audience.

While Web3 and blockchain continue to appear in the mainstream news, such products and services are likely to appeal to those in the startup and technology communities – which we know lack diversity, but which are also a relatively small slice of the global pie.

It is difficult to estimate what percentage of the world’s population works in startups. In recent years, the industry has created approx three million jobs in the US When measured against the total US population – and disregarding the lost jobs – the technology industry is nowhere near representative of working-age citizens.

Until Web3 solutions become more mainstream and broaden their appeal and use beyond those who have an inherent interest in technology and become affordable and accessible enough for a wider population, access to high-quality data in sufficient volumes to train AI systems will become an important remain an obstacle. The industry must now take steps to address this problem.

Alexandra Karpova is head of marketing at Lumerin.