By Cesar Tavares, Sr. Director of Technology and Innovation
Artificial intelligence (AI) and blockchain — they are two emerging technologies that, while powerful, are limited on their own. However, when put together, AI and blockchain have the potential to become an innovation force multiplier in the Federal Government. To better understand the value and potential of combining AI and blockchain in government, let’s take a close look at the relationship between these emerging technologies, then at the technologies themselves.
AI, blockchain, and the data connection
At first glance, AI and blockchain might seem an unlikely pair. AI and blockchain are fundamentally different technologies. AI tends to be a more centralized, focused, black-box solution that executes algorithms, while blockchain is a more distributed white-box value transfer trust mechanism. Whereas AI gives us the tools we need to extract value from data, it is blockchain that ensures the safe transfer of data assets. Though they are fundamentally different technologies, they both require data to operate: AI uses data to automate cognitive decision making, while blockchain uses data to build trusted business networks. This means AI and blockchain have a symbiotic relationship. AI needs trusted data to be effective, whereas blockchain has trusted data but does not have the autonomous sophistication of AI. The inherent weakness in one is the strength of the other. This makes for a perfect recipe — combine these two technologies, and you can create powerful Federal Government solutions like no other.
The AI challenge
With the advent of the internet came a proliferation of data, and our inability to process and extract value from all this data inevitably lead to the fields of data science and artificial intelligence. AI can ingest, categorize, label, and make inferences and predictions about the information processed. This, tied with the sophistication of neural networks, gives AI abilities that resemble and, in some cases, supersede human cognizant reasoning. (Remember 1997 when IBM’s Deep Blue beat the world chess champion Garry Kasparov? Think of how much more powerful AI systems are today!)
As indicated, AI was designed to make sense of data rather than transfer value. For example, when you send someone a digital file (e.g. email, photo, tweet, etc.) you are not sending the original, but rather a copy of that file. Now consider a scenario where this digital file has some intrinsic value. What if the file exposes your identity, legal documents, voting ballot, health records, currency, etc.? How would you feel about someone being able to duplicate the file and ultimately run off with your asset? Just this February, DISA alone exposed the personal data of about 200,000 people according to FNN.
In the past, the only way to ensure data immutability was to employ auditors to verify data integrity and validity. Unfortunately, this has the inadvertent effect of creating replication of data stores which lead to siloed organizations, heavy data reconciliation costs, and lack of synergy among teams. Worst of all, this approach diminishes the most important requirement for governments to thrive, which is trust.
Now imagine a world where the network has a trust mechanism already built in, aka blockchain. In this network, entities are able to exchange assets without the need for middlemen. That is the promise of blockchain — it is a secure value transfer trust mechanism. It works because blockchain transaction data is stored in encrypted structures called blocks. These structures are tied together in a chronological chain of events, and new blocks are only added when network participants agree on the validity of transactions by way of a consensus mechanism. Assets tracked within this architecture have the benefit of being encrypted and immutable and have provenance.
Whereas AI can help make sense of an asset, it does not do a good job of authenticating it. That job is better suited for blockchain because digital assets can be cloned, copied, replicated, and shared infinitely. This creates a predicament known as the “double-spend” problem. As the term suggests, double spending means you are at risk of spending the same asset twice and if left alone, this problem can lead to a dysfunctional network and potentially devalue any asset on that network.
The blockchain challenge
Yet, blockchains are not without their own challenges. In simplistic terms, a blockchain is a tamperproof recording device that never stops recording, and like any system that keeps getting bigger, it will eventually become vulnerable to exceeding its capacity. As block sizes increase, so does the computational power needed to process them; and as the number of transactions increases, so does difficulty in adding new nodes. This translates to a slower, more expensive, and unsustainable system where the only recourse is to fork. Fortunately for us, AI is here to help.
If we continue to use the recording device as a blockchain analogy, we will quickly realize that keeping the recorder on all the time is perhaps not the most effective mode of operation. A better solution would be to make a judgment call and limit what gets recorded. These are the types of problems that AI is well suited for. Here we could use AI to sift through the noise and make judgments about what data is worthy of being recorded onto the blockchain. Applying AI logic in this way could address the scaling problem that plagues blockchain networks.
Other areas where AI could help improve blockchain include its “smart contracts” feature. Specifically, a smart contract is a piece of code on the blockchain that allows a transaction to occur on an asset only when all conditions are met. Today, these smart contracts are still relatively simple; however, with the progress of artificial intelligence, it is possible to make them more powerful and capable. For example, we could use AI to analyze past contract behaviors, make suggestions, and predict which type of contracts would most likely secure a future agreement. According to Fintech Law, AI could be used to construct smart contracts through “shallow semantic parsing, named entity recognition, co-reference resolution, and a host of other techniques.”
Looking ahead
Today’s industry is powered by technology and fueled by data. Some have even gone as far as calling data “the new oil” of the fourth industrial revolution. If data is indeed the “new oil” then AI and blockchain can be compared to the new oil refineries. In this new world, the institutions with the most data that can leverage these technologies have the potential to gain the most. As it stands, there is no other institution that holds more data than the Federal Government.
The government is aware of this potential, which is why we continue to see significant investments in these areas. According to the FCW, for example, “market researcher IDC projects that federal government’s spending on AI will grow from $250M in 2018 to $1B in 2023. Bloomberg Government is even more bullish, with federal AI spending reaching $1.1B in 2019 following 43% year-over-year growth.” Perhaps even more important than the spending is the ROI projection. FCW’s model projects “a $364 billion gain in productivity for the federal workforce in 2028” given the projected investments.
Blockchain spending has also increased as noted in the IDC Government Insights report. According to the report, spending on blockchain will likely increase from $10.7 million in 2017 to $123.5 million in 2022. There has also been an increase in government blockchain prototypes as seen in HHS’s use case to secure and track network log files, FCC’s use case to track pharmaceuticals, the Navy’s use case to encrypt messaging, the IRS’ use case to track cryptocurrency, and many more.
Clearly, the opportunity to benefit from these technologies is recognized by our government. However, what’s not yet obvious is how combining the two could be a game changer. The AI-blockchain paradigm is a wide-open space for innovation. In Part II of this series, I will demonstrate how Octo implemented a blockchain solution for HHS and how this solution could be enhanced using AI.