A survey of 2,000 UK workers who had applied for at least one job in the past two years found that 75 per cent were unaware they have a right to challenge an automated rejection.
The right in question
The London non-profit People Like Us is running a campaign called Reject the Rejections, pressing employers to improve transparency and accountability in automated recruitment. Its chief executive Sheeraz Gulsher points to guidance from the Information Commissioner’s Office stating that people subject to certain automated decisions have the right to request a human review.
“We’re trying to remind people of that right, as many simply aren’t aware of it,” he said.
What he means by an “AI war”
Gulsher describes the present situation as an arms race: candidates “resorting to AI to bolster their applications” while “employers are dealing with huge amounts of applications” and using AI to sift them. Each side’s automation is a response to the other’s, and the volume grows on both ends.
His request of employers is specific and modest: audit these systems, publish the results, and have them independently assessed. “We can’t say for certain that every company is operating its automated or AI systems without bias.”
The finding that should stop you
Seventy per cent of ethnically diverse jobseekers said they had altered or concealed information on their CVs — including their country of origin, religion or name — out of fear that recruitment software would reject them. Among white candidates the figure was 57 per cent.
That is not a complaint about technology being inaccurate. It is people pre-emptively editing their identity out of a document because they expect a machine to hold it against them. Whether or not the software would have, the belief is already changing behaviour at scale — and a hiring system that quietly teaches people to hide where they are from has a problem that no accuracy metric will show.
Why this matters in Bangladesh, and differently
Bangladeshi applicants meet these systems from the far side. Applications to foreign employers, to remote roles advertised globally, and increasingly to large local employers using imported applicant-tracking software, are screened by tools trained mostly on Western CVs and Western career patterns.
A Bangladeshi CV often fails those patterns innocently: degrees from institutions the system has never seen, employment gaps that are normal here, names the parser mangles, dates in a different order. None of that is bias in the political sense. It is a model performing badly outside its training distribution, which is the same failure described in AI’s handling of Bangla, wearing different clothes.
The practical point is that the right to a human review mostly does not exist here. There is no Bangladeshi equivalent of the ICO guidance, and a candidate rejected by an overseas employer’s software has no forum at all. So the realistic advice is unsatisfying but true: write the CV so a parser can read it — plain headings, standard date formats, no text inside images — and apply through a human where one is reachable.
Related: the other side of the same arms race.
