Blockchain helps regain confidence in artificial intelligence decisions

< IMG SRC = "/Uploads/Blogs/8F/3D/IB-FS3AFBBBP6_BC9DCDEF.jpg" Alt = "Blockchain helps to return trust to artificial intelligence"/> ~ ~ ~ < P > About the growth of the role of artificial intelligence in making important decisions has been said, but the problem of trust in such decisions remains critical. Harvard Business Review has explored how companies can make algorithms more transparent and ethical & mdash; And it proposed an unexpected solution: the use of blockchain as a tool for trusting. We have prepared a brief statement of the most interesting of this analytics.

< H2 class = "" > artificial intelligence at business service & mdash; But not people

< p >Artificial intelligence & nbsp; nbsp; is increasingly used in banking, medicine, & nbsp; marketing, logistics, personnel policy. It analyzes large amounts of data, forms predictive models, helps make decisions in a second. However, users, customers and consumers often remain in the position of third -party observers: they do not understand how the AIs make decisions and whether these decisions are fair.

< p > one of the most acute problems & mdash; Lack of transparency. If a person has a refusal, release after an internal assessment, a wrong diagnosis or a biased decision, he rarely has the opportunity to check why the algorithm was behaved so. This is especially anxious in the context of racial, gender or social discrimination, which can be & laquo; sewn & raquo; into the model still at the training stage.

< h2 class = "" > Why current ways of trusting dont work

< p >Many companies are already implementing the mechanisms of ethical control of Shi & mdash; internal audit commissions, external checks, risk management policies. In some countries, regulatory bodies are working to require models certification. However, for the end user, this often remains unnoticed.

< P > In an attempt to maintain a competitive advantage of the company reluctantly reveal the details of the work of their models. Audits are usually confidential, and explanations of AI systems are formed in the form of general phrases or patterns. This further aggravates distrust.

< p >Another problem & mdash; The complexity of the system. Even if the company wanted to explain how the neural network works, it would be difficult to do in an accessible form. As a result, the user faces a black box that makes a solution with real consequences & mdash; without being able to influence, challenge or at least understand logic.

< h2 class = "" > blockchain as a bridge of trust between humans and algorithm

< p > in this context & nbsp; blockchain & nbsp; & mdash; Cryptocurrency technology, & mdash; It can become a tool of transparency and trust. It is not about storing the entire SI model in the circuit of blocks, but about fixing the key stages of its functioning in the unchanged, open to check the log.

< P > For example, the system can record in blockchain, which version of the model used, which input parameters are applied, what certificates this model has, whether it has been audited and when it was. You can also record the user's approval for data processing or his request to explain the solution.

< p >These records are not subject to reinstallation, and therefore create the basis for real accountability. The company will no longer be able to simply change the parameters of the postfactum model or hide the error. Blockchain serves as a digital trail that cannot be erased.

< h2 class = "" > Examples of implementation: health care & rsquo; i, finance, transport

< p > one of the most promising directions & mdash; Use of blockchain in the field of medicine. Imagine a situation where artificial intelligence analyzes MRI and makes a preliminary diagnosis. Recording what the model made the solution, what data it processed when the audit was held for the last time & mdash; All this can be fixed in the open registry. In case of error or appeal, this will allow to trace the chain of actions and attract the responsible.

< p >In the financial sector of companies that use CIs to score creditworthiness, they can retain data on certification of their algorithms, their compliance with anti -discrimination standards, training samples. The customer will be able to claim that he has not been discriminated against gender, age or place of residence.

< p > Another example & mdash; Management of autonomous vehicles. Blockchain allows you to record the key solutions that AI makes during the movement. In the event of an accident, you can reproduce the logic of actions, as some companies do in the US within the pilot projects.

< h2 class = "" > Why is it profitable business

< p > At first glance, these are only extra costs: technical integration, legal support, publicity of processes. However, in practice blockchain gives some important advantages.

< p >First, transparency, certified by independent technology, reduces the risks of regulatory sanctions. In the future, having such a mechanism may be a must & rsquo; a vocal requirement, especially in Europe.

< p > secondly, it works for reputation. Companies that demonstrate openness and readiness for inspections form a higher level of trust by customers, which is eventually turning into loyalty and long -term cooperation.

< P > Third, blockchain creates a convenient infrastructure for inter-organizational interaction. In cases where several partners & mdash; For example, an insurance company, a bank and a medical service platform & mdash; Transparent joint registry reduces the risk of misunderstanding.

< H2 class = "" > How to introduce: where to start companies

< p >In order for the use of blockchain not to turn into extra bureaucracy, experts advise to start by fixing critical points: for example, launching a new model, passing audit, changes in the logic of the algorithm. Not obliged to create a public blockchain & mdash; can introduce a private registry and then scalable.

< P > It is very important that the implementation is accompanied by thoughtful communication. The user should understand what is fixed, why and how it protects his interests. Platforms that integrate interfaces with Rosquo; Rsquo; Custom Actions AFS, get more positive feedback than those just adding item & laquo; consent to processing & raquo;.

~ > 62 > 62 < P > Gradually, such registers can turn into industry standards. Both at one time ISO-certification became a marker of product quality, and & laquo; register of trust in Shi & raquo; can become a marker of responsible technological company.

< h2 class = "" > transparency as a new norm by

< p > Artificial intelligence & mdash; Not only a powerful tool but also a serious & nbsp; a call of ethics & nbsp; and responsibilities. Technologies such as blockchain make it possible to turn a black box into an open system in which the user has the right to know how the decision is made. Companies that are the first to introduce such approaches not only minimize risks, but also benefit in the long run & mdash; After all, trust becomes the most valuable currency of digital era.