Old Eyes on ‘New’ Tech - AI & Blockchain
D Health captures the thoughts of one of the domain experts, Dean Mann from the start-up BlakesEye, who will feature in the next corporate working group meeting.
AI and I
For reasons probably best known to my genes I have worked balanced on the cutting edge of technology for nearly 4 decades. I first encountered A.I or Artificial Intelligence back in the early eighties. Using some of the early microprocessors and hand built memory arrays that took up a rack of electronics as tall as I was, we ‘taught computers to see’, at least that’s how I saw it at the time. We were trying to identify abnormalities (e.g. tumours) in the brain using 3D CAT (Computer aided tomography) scans. We call this Machine Learning and the idea of teaching a computer is valuable in understanding the challenges in realising value from AI today.
Simply put: Then, as now we have to train AI to interpret useful information from data. The AI we encounter today often utilises pre-trained models for example in image, speech, handwriting or facial recognition. Advances in computing power, data storage and battery performance mean that even our smartphones and their cameras can utilise powerful AI models especially if combined with computing in the cloud. It is important to note that using these models, the computer knows what a dog or a human face looks like having been trained on thousands of images with errors being checked and corrected to optimise the real-time response.
This is all very well if a business needs to utilise existing trained AI models. But what about training new models? What if we want to identify the tumour in the brain or a poorly performing branch store in a large business and a trained model is not available? The challenge will be the considerable time and effort required to gather suitable training data, train the model and adjust it to minimise the errors. Sometimes the internet can be used as a source of training data, but often suitable data is not easily available with the only source being from human expertise and experience with relevant data. In some areas, unsupervised machine learning can be applied to identifying clusters of data; this still needs to be interpreted in terms of the business or operational value. Although considerable work is being undertaken to evaluate the power of AI, for example, in healthcare for screening, it is important to appreciate the amount of data and human expertise in applied healthcare that is required to train a new model to be effective.
Realising the potential of AI and achieving an acceptable ROI (return on Investment) requires careful analysis of the problem to be solved, the applicability of existing AI models and any need to capture training data for new models.
Going around the Block
What about the new boy on the block or should I say Blockchain? Like AI, Blockchain appears everywhere technology is mentioned. Like AI it is a beacon of potential to solve the problems of the world. Like AI it is poorly understood and misrepresented both as saviour and hype. Realising the potential of blockchain has not been helped by some attempts to explain it; not because they are wrong, but because the analogies used generate further misunderstanding. Blockchain is sometimes referred to as a distributed/decentralised database, which is true, in a sense but we tend to view databases as the management systems that allow us to Create, Read, Update and Delete data via applications such as SQL Server, MySQL, MongoDB, even Excel etc. Data stored in a blockchain can be created and read but not directly updated or deleted. So what use is it and why the hype and interest worldwide?
Blockchain technology stores data first as an unchangeable or immutable ledger. This is its core value. Secondly this ledger is decentralised, no single entity validates the ledger, instead this is done via consensus between entities on a network. Think of the blockchain like an accounting ledger. If an amount is entered into the ledger and later has to be changed, this is done by an additional entry in the ledger, not by deletion and replacement of the old record, both entries are still present in time sequence. So this approach embeds a sequence and history for the data stored. If the entries are timestamped, then a more useful timeline is established - an auditable trail. If the integrity of the blockchain can be assured, the ledger can be used in exchanging value as a currency, or in tracking the movement of goods like pharmaceutical drugs, vehicles or other assets. It can hold a patient record or details of transactions as part of a contract between individuals or organisations. But this can all be achieved using ordinary databases with certain rules applied. However, all the potential applications mentioned require considerable trust in the integrity and security of a database often in the hand of a third party. The difference comes down to how, without a trusted third party blockchain achieves integrity and trust even where none or very little exists.
For parties in a contract, trust is usually secured by payments, deposits, credit history and the like, relying on third parties like banks or governments, trusted to provide guarantees or to manage payments upon completion of elements of a contract. The proliferation of intermediate (middleman) services extracts significant value from the contract process and applies influence aligned to their interests rather than that of the parties at either end of the contract. The blockchain is potentially disruptive because it can eliminate or restrict many intermediate services associated with these existing methods of trust. It is exercising many service industries like banks and insurance companies along with national governments in evaluating both its potential and threat.
The blockchain decentralises the ledger. Every entity involved in a contract can hold a copy of the ledger and can confirm the validity of the copy they hold via mathematical consensus mechanisms that do not rely on trusting anyone else. By smart use of mathematics, encryption, game theory, networking and the internet, the blockchain has enabled the Bitcoin cryptocurrency to achieve tradeable value against more traditional (fiat) currencies. This is without Bitcoin being controlled by any government or other single entity. The ability to establish currency and move money from peer to peer without banks, beyond individual governmental control is enough to draw the attention to blockchain technology alone. However, it is the ability to redefine the nature of trading and contractual work, peer to peer, that has the greatest potential to impact all our businesses and organisations. Shifting control of the supply chain from the middle to the ends connects the manufacturer and the ultimate customer directly and increases their combined influence and control of the process. It can eliminate unnecessary intermediate services and distill down to the essential ones for the contract and open those services to a wider range of suppliers.
I think there is opportunity for every business and the blockchain at the appropriate scale. Not every business will benefit from worldwide networks and cryptocurrency trading, however, many can benefit from blockchains and smart contracts on private trusted networks but the technological implementation still needs development to capture the mainstream of our trading activities.
I, AI, BlakesEye and the blockchain
In my company BlakesEye, we are developing our own small footprint hybrid blockchain applications to provide asset operations and management for companies in renewable energy, Oil & Gas exploration and decommissioning. The application and technology will enable significant improvements in high value asset up-time and availability. In healthcare it will have direct application in the pharmaceutical supply chain and potentially for the security, ownership and control of the patient record. We also see exciting potential for the blockchain and AI in combination. The old eyes are still working with the new tech. With experience, there is real value from the cutting edge
BlakesEye Ltd 2018
22 views0 comments