Despite all the guarantees and checks, you can often come across fraudsters on the freight market. Svyatoslav Vilde, Director of the ATI.SU Freight Transport Exchange, talks about how artificial intelligence fights cybercriminals and determines fair prices for logistics services.
Risks of logistics companies
The freight market is very dynamic. New carriers are constantly appearing, customers of transport services are looking for the most profitable contractors, rates for services are changing under the influence of external factors. As a result, problems related to fraud and non-transparency of processes remain relevant even despite digitalization.
One of the most common fraud options is that the intermediary receives money from the client, but does not pay the contractor. In a few weeks, he picks up many such orders, and then disappears with the funds received. Frequently, the loss of goods also happens – the carrier may simply not deliver the goods to the recipient.
There are several reasons that do not yet allow to completely eliminate fraud, even on digital freight transport platforms:
- New players are constantly appearing whose reliability cannot be verified initially. Some of them even operate as respectable firms for a while;
- Fraudsters buy up the accounts of trusted carriers or forwarders, sometimes together with a legal entity and SIM cards required for identification in the system.
Artificial intelligence in logistics
AI in logistics industry is still used to a limited extent, market participants are still studying its possibilities. For example, we use a machine learning algorithm to analyze the activity of platform participants. A special program compares information about them with the data specified in the accounts of companies, and identifies inconsistencies.
A simple example: a conscientious performer changed the field of activity, and scammers bought his account on the platform. Artificial intelligence should “understand” that this user was inactive for some time, and then began to act in a different way. Over time, the learning system finds similarities and forms behavioral models of both bona fide participants (forwarders, freight carriers, shippers) and malefactors. Such nuances as an abnormally high rate or a long absence and a sharp change in user behavior on the exchange serve as an alarm signal and a reason for additional verification.
AI also helps companies “predict” the fair market value of shipments. Based on the data accumulated in the past, the algorithm determines which price is most likely to be set by the transport company in the future. If there were a lot of information in each direction, then it would be possible to get by with simple statistical methods and not use complex models and machine learning technologies. Read more on https://data-science-ua.com/
Long-term game, game for the result
So far, the introduction of artificial intelligence technologies into the processes of logistics platforms is rather slow. This is due to several reasons:
- In order to competently train AI, a large amount of relevant data is required. Unlike, for example, retail or banking, where they have long relied on Big Data, logistics companies have not yet accumulated enough information that is needed for the digital algorithms to work correctly.
- Among the participants in the logistics market, a clear and unambiguous idea of what kind of data and by what criteria should be considered has not yet formed.
- There is no practice of developing and implementing solutions using AI – ready-made customized analytical modules.