The generative AI genie is out of the bottle, but how do we harness it?

A fundamental concern is the risk of embedded bias. 

While many people think of generative AI as some all-knowing mind, the technology’s large language models (LLMs) are more like giant libraries. AI tools give almost-instant access to a vast bank of information, published over the last century and drawn from the internet. As they are trained on historical data, the results they generate can reflect the biases and distortions already present in that data. There is a real risk of conflict with today’s values around equality, diversity and inclusion (EDI).

The bias problem in recruitment

The main use of AI in recruitment is screening applications for high-volume hiring. Back in 2018, Amazon abandoned an experimental AI recruiting tool after it showed bias against women. It had been trained on the historical CVs of its male-dominated tech workforce. Another case involved video interview platform HireVue. It removed its facial analysis capability after an audit suggested it could disadvantage candidates with disabilities, non-standard accents, neurodivergent behaviours or other traits.

Developers may claim their latest AI tools are impartial, but research has shown that AI can amplify discrimination due to unrepresentative training data, skewed feature selection, and hidden proxies for protected characteristics – such as education history, employment gaps or postcodes. 
The ‘black box’ nature of AI can also make decision-making more opaque. And AI bias is more than just a data problem: a model can learn and perpetuate pre-existing bias embedded in an organisation’s processes and past decision-making.

A double blow for entry-level workers

There are particular reasons to worry about the impact of AI on entry-level jobs. 

Not only is there greater risk of biased AI screening for these kinds of roles where applicants have limited experience and rely on transferable skills (with applicants filtered in ways that employers don’t understand and candidates cannot challenge). There is also growing evidence that AI is shrinking entry-level opportunities in many sectors by automating the tasks those jobs involve. 
A study by King’s College London reported last year that firms at the forefront of AI adoption reduced total employment by 4.5% on average, with junior roles falling by 5.8%. The World Economic Forum concluded that AI could already handle 50-60% of typical admin tasks undertaken by junior staff (report drafting, pulling research together, scheduling, cleaning data) and warns of an ‘AI tsunami’ sweeping away youth employment opportunities.
Some employers will be tempted to slash these jobs so they can reduce costs and increase efficiency. But taking away the traditional bridge between education and employment is short-sighted and harmful. In any organisation, juniors ‘learn the ropes’ by grappling with the everyday details of operations and procedures. That experience provides the foundation for learning, career development and more senior roles. 
Cutting back the junior ranks will mean less rising talent and fewer future leaders. AI may provide more consistent and predictable results compared with raw recruits learning on the job. But can it generate the corporate knowledge, insights and innovations that we imperfect, yet ingenious, humans produce? 
Displacing these jobs will erode the richness and diversity of multigenerational workforces. And the wider impact on society could be profound, driving up the already worryingly high number of young people in the UK who are not in employment, education or training (NEETs).

Risks require safeguards

Those are just some of the very real risks from overly hasty adoption of the technology. However, it cannot be un-invented – the GenAI genie is out of the bottle. Both recruiters and candidates are already embracing AI in the UK. More than half of jobseekers (51%) said they’d noticed AI being used during recruitment in a 2023 survey, while almost as many candidates (46%) used AI themselves when searching and applying for jobs.

Companies, including those in the waste management sector, must harness technological advances, including AI, to keep our people and the public safe, remain competitive, drive continuous improvement and serve customers better. But we need the right guardrails to ensure we protect equality, diversity and inclusion. 

Using AI responsibly

So, what’s the responsible way to harness AI in recruitment? 

At SUEZ in the UK, we are currently using AI to assist us in refining the wording of advertisements to ensure they are inclusive. These are reviewed for compliance with our EDI guidelines. The technology also informs independent analysis of our employee engagement surveys. But we continue to explore where AI can genuinely help improve our HR procedures and processes, as well as improve performance in other areas of our support functions and wider business. 
Where employers use AI in the staff selection process, it should only support human decision-making – not replace it. Humans must remain responsible for review, exceptions and final choices. This is in line with government guidance which calls for assurance before deployment, clear governance and ongoing monitoring.
Reviews should include checks that the AI application is only scoring skills and behaviours that are genuine criteria for the role. Regular audits are needed to guard against bias.
It’s crucial that recruiters and managers are trained to understand AI’s potential but also its limitations, common failure modes and legal risks. The Chartered Institution of Wastes Management (CIWM) offers a free course for members covering the ethical considerations around AI, and those of us in the CIWM’s Sector Inclusion Forum will be reviewing the EDI aspects. 

Can AI transform our business?

I am excited by new technologies and can see how AI-powered tools may help us streamline internal administrative processes, identify patterns and risks with the vast amounts of data we are collecting, and lead to smarter and safer working in operational areas. It could help with route optimisation for our vehicles, materials sorting, and identifying non-conforming waste. It could help identify hazard out on the roads and streets where our vehicles are operating.  

And in HR too, we need to keep an open mind on AI’s potential. Rather than introducing or adding to bias, the technology might even help some employers reduce it. There are AI tools designed to detect hidden patterns of discrimination.
AI may be able to improve candidate matching by enabling more comprehensive skills-based assessments that look beyond conventional CV criteria. Perhaps AI-enhanced analysis could become better at identifying people with strong transferable skills even if they lack relevant experience or a ‘typical’ background?

Redesigning and enriching roles

And what about those jobs most vulnerable to AI replacement? There’s an urgent need to promote retraining and upskilling, as well as the redesigning of roles. 

As repetitive tasks such as data entry and routine customer enquiries are automated, staff can focus more on exception handling, analysis and other higher-value activities. That matters in an industry that can struggle to attract talent until the variety and value of the work we do is better understood. However, jobs involving AI-enabled workflows could attract more young people. Training people in traditional and/or transition roles to work with AI in new applications that transform our business would be a logical next step for our in-house Learning and Development Team and the Digital and Data Academy they have created in partnership with our IT function and the wider business. 

The challenge is to exploit the technology and govern it responsibly, balancing artificial and human intelligence and widening opportunity, not narrowing it.
My own hesitancy stems from the fundamental concern that we must not build the new structure of work on the flawed foundations of employment practices in the 1980s and before. We must not go back to create the future. Instead, we must always ensure the evolving world of work we help to shape is free of the inequities, disadvantage and harm caused in the past.