News / Don’t be fooled – logistics AI uses few and far between, warn experts

© Stnazkul |

For logistics, AI risks becoming another ‘blockchain’ – a shareholder-magic-dust with little near-term practical application, according to experts.

In a version of the wider cryptocurrency ‘pump-and-dump’ ecosystem in miniature, during the peak of the hype, investors were beguiled by marketing materials claiming companies were ‘doing something with blockchain,’ though its actual uses were seldom identified. This attracted huge investment without firms having to follow through on their claims.

“A child born out of the blockchain hype seen five years ago” – as described by Vespucci Maritime’s Lars Jensen – Maersk’s TradeLens, an attempt to implement a blockchain-based network in earnest, folded last year. TradeLens had garnered many big-name ‘trust anchors’ and ‘foundation carriers’ – companies which would, in effect, co-sign the technology without having to invest. But few actually signed up as customers.

“For ECMRs (electronic consignment notes), electronic bill of lading for containers, proof of delivery – in that sense blockchain makes sense,” Sarunas Belickas, chief operating officer at Cargo Stream, told The Loadstar. “But [TradeLens] was too advanced, and didn’t actually solve any real problems. Considering that SMEs make up 98% of the logistics sector, a logistics professional could not begin to understand what it is supposed to do, from that perspective.”

Amid news that companies are tripling their share prices by pledging to do something with AI, it looks like we are on the verge of yet another ‘blockchain’. “I don’t think we learned our lesson from blockchain,” Mr Belickas said. “We see evidence that investors are throwing money at anything that says ‘AI’ now.

“These are neural networks trained on data-sets, technology which has existed for some time. But it is also technology that is even harder to understand [than blockchain]. Investors will be attracted by this.”

Already, scammers are selling courses on using AI – akin to a thousand-dollar tutorial on how to google. Nonsense claims abound over what the technology can do, including the assertion that ChatGPT could start and operate an entire business without human intervention.

These claims obscure the limited, but very real, utility that lies beneath the hype, Mr Belickas explained. “You really could not go to step B: putting management of essential business processes in the hands of an AI, before figuring out step A: booking transportation with email without TMS, using AI.

“But there are use cases — for example, it could improve visibility,” he said. “People will perhaps not need to scroll and review every order – an AI could do this, say in a case where a shipment’s ETA has increased. If it is a dramatic case, the AI could identify this and draw the user’s attention to it – or, dismiss it if it is not a major concern.”

Another example is in communication. “Perhaps not every bill of lading needs to be filled out manually. It’s possible that an AI could do this. We are planning to use AI for communication.” 

The Loadstar ran this article through ChatGPT, a language model developed by OpenAI. Instructed to act as editor, correcting any mistakes, it found 10 serious grammatical errors – none of which were in the original text. On pointing this out, ChatGPT was extremely polite.

“I apologise for the confusion and any frustration caused,” the chatbot replied. “As an AI language model, I should have been able to accurately identify any errors in the given text. Upon reviewing the given text again, I did not find any examples of spelling or grammatical mistakes in the provided text.”

If only every editor was so magnanimous.