Verifiable credentials in the age of AI

It feels like barely a moment has passed from a chatbot clumsily trying to help us support any website, to seeing a chat like GPT-4 who, remarkably, passed the standardized exam to become a lawyer in the United States.

The accelerated growth of AI has been so impressive that more than 1000 leaders, such as Elon Musk, Yuval Harari, Steve Wozniak, among others, have signed a letter requesting a pause in the development of the models, while safety protocols and audit (although this is unlikely to happen).

As human beings, we are not prepared for abrupt fundamental changes, and that is precisely what this technological paradigm shift represents, of which we have not yet fully understood all its implications in the world of work and in the relationships between individuals.

Early forecasts relate to tasks that used to be performed by humans and are now more likely to be automated at a fraction of the cost, which has implications for role definition and the need for in-depth reskilling for much of the workforce. A solution based on GPT-4 or another similar model can already read a 200-page legal contract in a fraction of a second and, from there, identify gaps and recommend clarifications. It may not eliminate all the lawyers in a company, but if 3 were needed before, now only one will be needed.

AI and human behavior

A new concept emerges: AI as a “co-pilot” rather than a replacement. Even several solutions already use this term in their products, such as Github Copilot or Microsoft Copilot. Star Wars predicted it, without R2D2 as Luke’s co-pilot, the story would be different.

Another possible outcome of this advancement in multimodal AI is the proliferation of superbots on social media. This is not new, but what is new is that they will be able to interact and impersonate human beings, deceiving the algorithms of these platforms (which, in turn, are other AIs) and, above all, deceiving real human beings. on those platforms.

This attack is based on exploiting a human characteristic called “herd mentality” or conformity bias, which has helped us survive as a species for thousands of years. It refers to our tendency to imitate the behavior and thinking of a group when we perceive a strong consensus on that action or idea. In other words, people do and believe certain things simply because many other people do and believe those same things.

Everyday examples of this are when we enter a restaurant because it is full or when we see several people looking at the sky and we can’t help but do the same.

The evolution of AI

The rapid evolution of AI has far outpaced the development of digital identity frameworks and trust systems. In fact, the owner of Twitter sees the only alternative as paying to obtain verification, in order to avoid the avalanche of bots that will flood the social network and cause all kinds of harmful effects on global conversations. However, what you don’t realize is that this would also come with its own drawbacks, as it would be a centralized identity that doesn’t belong to the users, and by setting a price to stay verified, you are creating a barrier that prevents people without resources. participate in that space (which may be legitimate in a private company, but not desirable in a social network that aims to be a platform for public conversation).

It might interest you: Why hire Acreditta instead of building your own development

No alt text provided for this image However, we are not condemned to always depend on Twitter, Facebook or Google for our identity in the digital world:

A new digital identity standard emerges

A few years ago, an open standard called “verifiable credentials” began to be developed, led by the W3C (World Wide Web Consortium) organization and the DIF (Decentralized Identity Foundation). These verifiable credentials bring to the digital world the layer of trust and identity that we have resolved in the physical world with diplomas, driver’s licenses, passports, DNIs, certificates and other elements that demonstrate our capabilities or characteristics validated by third parties.

These verifiable credentials allow us to bring all of these use cases (and some still unexplored) to the web environment, with technologies that improve physical credentials in several ways:

• Cryptographic security: They are impossible to falsify or alter thanks to verification based on blockchain technology. Most solutions even add a hash to a decentralized network as an anchor to provide new functionality to the credential.
• Privacy: People are the custodians and owners of their credentials, and the data they contain is not on any public blockchain (only a hash is used as an anchor, at most).
• Zero Knowledge Proof: Another privacy feature that allows people to submit only the relevant credential information or even submit a proof without revealing any of the original data.
• Instant Verification: Any third party wishing to verify the validity of one of these credentials can do so without needing to contact the organization that issued it, as the cryptographic proof is built into the verifiable credential itself.
• Portability and interoperability: Being an open standard, digital wallets are already implementing this format. This way, the person who owns the credential can choose which wallet to use or where to store it.

Verifiable credentials in the age of AI

The European Commission has already adopted this standard for its university diplomas and country identities, and the OECD is recommending its member countries adopt appropriate governance over digital identity.

AI has enormous potential to benefit humanity, but it can also pose complex challenges like those mentioned above. In both scenarios, verifiable credentials can play an important role:

In the case of “AI as co-pilot” that automates tasks previously performed by people, the need to acquire new skills requires capturing that knowledge in the form of micro-credentials or digital badges in a more detailed way. To handle the speed at which this can happen and the volume of people who could lose their jobs, it is necessary that new knowledge can be easily read by recruiting systems and verified instantly using the verifiable credentials format. Current human verification through phone calls with institutions is not viable in this context.

In the case of AI impersonating humans on social networks, one solution could be to prioritize the interactions of users who present their ID in verifiable credential format. For this, it is necessary for countries to follow the example of Europe and the recommendations of the OECD. Alternatively, private companies could emerge that perform an identity verification (KYC) process and issue verifiable credentials that function as proxies for a unique verified identity.

To complete…

Beyond these scenarios, this open standard offers exciting possibilities and use cases that put the power back in the hands of the people and balance the scales in an era where personal data has been used indiscriminately without regard to the owner of that data. Everything we achieve in our academic and professional life is part of our identity, and since identity is a human right, we must pay attention to these protocols that will strengthen our way of relating in the digital society.

Subscribe to blog here and receive all our content