Should AI judge who gets a mortgage, loan or phone contract?
BBC Technology is asking whether artificial intelligence should help decide mortgages, car loans and mobile contracts, putting automated judgment at the center of everyday credit decisions.

In this piece
A BBC Technology report raises whether AI should help judge applications for mortgages, car loans and mobile phone contracts. The question lands as lenders…
BBC Technology is asking a pointed question: would artificial intelligence be a good judge of whether someone should get a mortgage, a car loan or a mobile phone contract? The item frames creditworthiness as a decision that can shape housing, transport and basic connectivity, and it asks whether a machine should sit in that judgment.
The report itself poses the issue rather than settling it. It does not, in the material available here, set out a new product launch, a regulatory ruling or a measured error rate. What it does is place AI next to three ordinary forms of credit and ask whether that pairing is sound.
Lenders have long sorted applicants with rules, scores and automated checks. Those systems already influence who is offered a loan, at what price, and who is turned away. AI is often described as a further step: models that look across more signals and update as patterns change. Whether that step improves judgment, or simply hides it, is the dispute the BBC headline puts in front of readers.
A mortgage is usually the largest borrowing a household will take on. A car loan can determine whether someone can reach work. A phone contract is smaller in cash terms but still a gate to banking, job searches and public services. Treating all three as candidates for AI review underlines how widely automated credit decisions can reach.
Supporters of data-driven underwriting generally argue that consistent models can apply the same criteria to every file and can spot risk that a rushed manual review might miss. Critics answer that models learn from past lending, so old exclusions can be repeated at scale, and that a refusal is hard to challenge when the reason is buried in a score. Both lines of argument are familiar in the credit industry; this report does not supply new evidence that decides between them.
Accuracy is only part of the test. A system can be statistically useful and still be unacceptable if applicants cannot understand a decline, correct a mistake or appeal to a person. Credit files contain errors, and people change jobs, addresses and family circumstances. Any AI judge would still depend on the quality of the data it is given and on the policy choices built into it.
Accountability is the other open problem. If an application is refused, it is not always clear whether the fault sits with the lender, the model builder, the data supplier or the rules the firm chose to encode. The BBC piece asks whether AI would be a good judge; it does not identify who would answer for a bad one. That gap matters because credit decisions are not advisory. They allocate money and, in practice, opportunity.
There is also a difference between assistance and authority. A model might flag risk for a human underwriter, or it might approve and decline on its own. The public question in the headline — should AI be judging creditworthiness — is sharper if the system is the decision-maker rather than a tool on a desk. The available report does not say which of those designs is in view.
Regulators in many countries already expect lenders to treat customers fairly and to be able to explain adverse decisions. How those duties apply when a model is complex is an active policy debate, not a settled standard described in this item. Firms experimenting with AI in credit will still have to show that outcomes are lawful and that people can contest them.
What happens next is not spelled out in the source. Readers are left with the question itself: whether an automated system should help decide mortgages, car loans and phone contracts. The answer will depend on evidence that this short report does not provide — how such systems perform, how errors are fixed, and whether a person remains responsible when the model says no.
Primary source: BBC Technology. Newzly rewrites and structures source reporting for clarity; verify critical details at the original.