CEO's Blog: What AI means for the executive search industry

New technologies are almost always over-hyped. Yet it seems likely that the arrival of powerful new Artificial Intelligence (AI) tools over recent months might be a rare exception to that rule. This tech clearly has the potential to start a paradigm shift on par with the creation of the internet, or perhaps even the invention of the printing press. Such shifts are rare in human history. They affect so many complex systems that they invariably change our relationship with information, and with each other, in ways that cannot be predicted at the outset. Given all this, making grand predictions about “what AI means” for any industry is probably foolish at this early stage. Nevertheless, I feel we have to start somewhere, so I wanted to explain our initial thinking on this topic for the benefit of Society’s various stakeholders.

Let’s start with how we are using AI already. For the past few months, colleagues within Society have been experimenting with tools such as ChatGPT, QuillBott, and Jasper. We have had demonstrations and training sessions internally, and lots of conversations to share our discoveries and to uncover best practice. Right now there are five key ways in which it’s starting to make itself useful in our day-to-day work:

1. Getting a first draft moving. Since it is primarily LLMs (large language models) that have been at the forefront of the AI revolution, it shouldn’t be surprising that their biggest impact so far has been on our production of written content. When a colleague has to come up with 300 words for a Candidate Information Pack on "the best things about Lahore as a location" or "the cost of living in San Francisco", it can take time to decide where to start. Tools like ChatGPT are helping colleagues to break free from writer’s block or ‘the tyranny of the blank page’, by providing a few initial ideas that can then be built upon, expanded and finessed. This speeds up the drafting process considerably.
2. Concierge service. Similar to the above point, we are able to use AI to generate ideas. For example "Recommend five books about neurodiversity" or "What's a good location for a discreet business meeting in Lower Manhattan?"
3. Simplifying and condensing. Often we have to take complicated plans and concepts and distil them into messages that can be conveyed to a wide variety of different audiences in a challengingly short period of time. Asking an AI tool to digest something and "Summarise this text" or “Explain this in terms a fifth grader could understand” can be a useful exercise that helps to zoom in on the truly salient points, ensuring that we're able to communicate as clearly and concisely as possible. 
4. Finding more examples. If we identify a target organization that looks like a strong source of potential candidates, or a couple of events or conferences where suitable candidates appear to convene, AI now provides our researchers with the ability to say “Find more examples like this” and quickly uncover additional possibilities that we might otherwise have overlooked.
5. Language translation. This is probably self-explanatory. For example, "Please translate the following job advertisement into Portuguese"

In just seven to eight months, these approaches have begun to deliver meaningful results and even started to feel routine. Over time, we anticipate that AI tools will have even more far-reaching applications. In particular, we are excited about their ability to unlock hidden meaning in our data – spotting trends, making sense of patterns, and offering up new market insights for our clients (eg. around candidate compensation). However, there needs to be clear limitations and guard rails for something this powerful and untested. With that in mind, we have settled on four rules (although perhaps we should call them ‘Laws’ in tribute to the late Isaac Asimov), for how we will not be using AI. We consider these essential in order to protect the long-term ethos and integrity of our business:

1. We’ll continue double-checking everything. AI is far from infallible. Although the accuracy of these tools will no doubt improve over time, we will ensure that we still verify whatever AI tells us.
2. We won’t let AI communicate directly with clients or candidates. With the possible exception of scheduling meetings, contact from Society will always be from a real person. The authentic human touch must permeate all our interactions.
3. We won’t let AI make decisions about which candidates get progressed and which do not. For volume recruiters, some level of AI involvement in sifting and/or selection might become the norm. But at the level of seniority at which Society works it simply is not necessary or justifiable.
4. We will never input confidential or personal data into a public AI tool. Our commitment to candidate confidentiality and data privacy will remain sacrosanct.

In short, our ‘AI philosophy’ is this: "We will use it to augment rather than to replace existing capabilities. It will only ever be a digital intern or co-pilot. We will resist putting it in a position of control."

Even with those limitations though, this tech will still transform our industry. Can AI replace executive search in its entirety? No, I don’t think it can. But it certainly has the potential to automate or commoditise parts of our work that were previously considered valuable. Such services include ‘removing the administrative headache’ of recruitment, ensuring that processes are sufficiently robust and well documented, creating materials that marshal and communicate core information about the client organization and the job opportunity, and perhaps even 95% of potential candidate identification.

If that’s so, what remains for the search firms? To my mind there are four areas that will be extremely hard for an AI to take over:

1. Cultivating trusting relationships. Even the most powerful tech won't easily supplant 300,000 years of socialisation. Individual candidates will still respond to the advocacy and persuasiveness of a fellow human being. That is particularly true when it comes to negotiations, where human give-and-take can sometimes secure outcomes that an algorithmically opimised solution would deem impossible or irrational. 
2. Gauging interpersonal skills and culture fit. Powerful and predictive psychometric tools have existed for decades, but clients still value a headhunter's human-to-human perpective on how candidates came across and interacted with them. It seems unlikely that AI will remove that preference.    
3. Diversity, Equity, and Inclusion auditing. Since AI systems feed on data that often has historical biases and forms of under-representation encoded within it, they remain worryingly susceptible to replicating those errors in their output. The nature of this tech is that not even its creators can fully understand or control how it works. As such, a need will emerge for third parties who can interrogate their outputs, spot their weaknesses, and create complimentary processes that attenuate their potential blind spots. 
4. Lateral thinking. Whilst AI might quickly master the task of aligning square pegs with square holes and round pegs with round holes, a human touch should still be able to add value through making more innovative and creative suggestions, both to candidates and to clients. Humans can occasionally see potential in people that even they themselves might not yet know they have.

Society will therefore prioritise investment, in terms of training, innovation and hiring, in these four areas. In doing so, we hope to future-proof our business, and to bring about a future that maximises the potential benefits of this astounding new technology, whilst controlling for and mitigating its potential hazards.