Alas, blockchain still is subject to human programming slip-ups

Time for more artificial intelligence in distributed ledger technologies, researchers urge.

Many see blockchain and other distributed ledger technologies (DLTs) as existing out in the distributed network, free from tampering, misdirects, and other human foibles. However, at its core, there are still human programmers needed to make it happen, opening up possibilities for loopholes and flaws. Artificial intelligence may help reduce such issues.

building-hannover-messe-march-2016-photo-by-joe-mckendrick-1.jpg
Photo: Joe McKendrick

That's the takeaway from a paper just published by Tshilidzi Marwala and Bo Xing of the University of Johannesburg, noting that the joining of blockchain with AI could help deliver an even smarter blockchain, which the authors describe as "Blockchain 2.0.".

The smart contract aspect of blockchain is a compelling market innovation. However, Marwala and Xing point out, "the defined standards regarding smart contract security is still lacking, which means the hidden vulnerabilities found in new or existing smart contracts could potentially lead to undesired results, e.g., monetary losses." Bugs in cryptocurrency smart contracts indeed have cost people millions of dollars.

Adding to the potential challenge, smart contracts in blockchain or other DLTs have a great permanence to them that may be hard to correct. "Unlike its counterpart centralized systems, once a smart contract is put into practice on a decentralized blockchain, rollbacks and compensations are often hard to be performed when coding errors occur," the researchers state. "Though human security audit could be a solution for smart contract, for creators lift their level of security, the cost associated with this practice tend to be a discouraging factor."

AI can help smooth the implementation of blockchain and other DLTs in the a number of areas, Marwala and Xing believe. AI includes capabilities such as formal verification that employs mathematical reasoning for debugging and learning, search-based software engineering and security. They see possibilities in the following areas:

  • Sustainability: "AI can optimize energy consumption" - a great concern now seen in the cryptocurrency space, as a lot of server horsepower is needed.
  • Scalability: "AI can perform collaborative learning without centralized data sets."
  • Security: "AI can detect blockchain applications layer intrusion issue."
  • Privacy: "AI can improve the performance of a hash function."
  • Efficiency: "AI can predict the likelihood of a node to fulfill certain mining tasks."
  • Hardware: "AI can enhance the design of mining hardware for an overall elevated performance."
  • Talent shortage: "AI can form multi-agent system for generating virtual distributed ledger agent."
  • Data gatekeeper: "AI can help with intelligently open data."

While there are some interesting demonstration projects taking place with blockchain outside the cryptocurrency realm -- such as the shipping supply chain initiative undertaken by IBM and Maersk, the food chain demonstration seen at Walmart, and even for insurance policies) -- there is still plenty of hesitation among enterprises to put their data assets out into a DLT environment. AI may help ease some of those concerns.