ChatGPT – Will AI replace humans in I&D training creation?

Raquel Silvestre, Senior Inclusion and Diversity Consultant, explores whether ChatGPT and Artificial Intelligence (AI) can replace humans in D&I training creation and discusses the value of human lived experience. The blog also alludes to the concerns of AI development in endorsing inequities, bias and stereotypes and the call for ethical, unbiased, representative, responsible and accountable AI development.

The rapid progress of digitalisation, automation and Artificial Intelligence (AI) is pressing us to consider and reconsider how we relate to it and the implications of such progress in the work landscape. One of the questions that Inclusive Employers received, which I will address here, is whether ChatGPT could replace humans in developing and delivering Inclusion and Diversity training. As someone introduced to ChatGPT not long ago by my younger brother, I’ll start by sharing my growing understanding of it.

What is chatGPT?

ChatGPT, or “Generative Pre-trained Transformer”, was created by the United States of America’s artificial intelligence company OpenAI. It utilises Natural Processing Language (NPL) like other conversational AI such as Apple’s Siri, Amazon’s Alexa and Google Assistant. However, Chat GPT also utilises Machine Learning (ML) algorithms that involve training artificial neural networks to replicate the complexity of the human brain generating human-like responses.

It is characterised by its capacity to capture the nuances and subtleties of human speech, build on previous conversations, and generate meaningful conversations with humans.

Meaningful conversations generated by ChatGPT means the capacity to understand the context and generate responses that reflect the user’s personality, mood, and intent.

ChatGPT can be used for conversational bots and customer service to generate creative writing and content.

In HR, for example, IBM is using a Chatbot named “Watson Recruitment Assistant” to improve their recruitment process (to answer questions, provide feedback, and assist with the scheduling of interviews); Unilever is using ChatGPT as “U-chat” to provide real-time responses to candidates, answer questions, and guide them through the recruitment process; McDonald’s is using a Chatbot named “McBot” to assist with their employee onboarding process (answer questions and provide training material)[1].

ChatGPT is trained on a dataset of 175 billion parameters and 570 gigabytes (GB) and can perform various tasks in different fields and industries. This is huge! To give us an idea, 570GB is equivalent to over 385 million pages on Microsoft Word[2].

As we speak, this capability has already progressed as ChatGPT-4 launched in March 2023. ChatGPT surpassed 1 billion users in the same month, an increase of almost 50% since the previous month[3]. Making it the fastest-growing consumer app in history. It might also be relevant to acknowledge that worldwide 11.72% of users are in the United States of America, followed by 6% in India, and an increasing number of nations have banned access to ChatGPT: China, Cuba, Egypt, Italy, Iran, North Korea, Russia, Syria, Ukraine.

No wonder this is a current hot topic as everyone is trying to understand how ChatGPT can help them in making life easier and work more efficiently. So, let’s now go back to the question that lingers.

Can AI like ChatGPT replace humans in writing inclusion courses and training programmes?

Curious to hear what Chat GPT would say, I asked, and this is what I got:

‘Yes, AI like chat GPT can replace humans in writing inclusion courses and training programmes. AI has the capability to generate content with a high degree of accuracy and speed, and can be used to create content that is tailored to the specific needs of the user.’

It continued to say that the training provided by AI is inclusive, relevant and effective. Out of curiosity I also asked: In what ways AI cannot replace humans? And surprisingly it also gave me a comprehensive answer:

‘Chat GPT is a powerful tool for generating text. However, it has its limitations. While it may be able to generate text that is grammatically correct and coherent, it cannot replace humans in writing inclusion courses and training programmes. These courses and programmes require a deep understanding of the subject matter and an ability to understand the context of a given situation to create effective programmes.’

Case closed, or so I thought. From this simple interaction, I could conclude that Chat GPT can in fact be a powerful research tool, collating vast amounts of existing data and relating this data to the user in a way that is relevant to what they search, hence meaningful. However, its potential seems limited by our capability to ask questions. Therefore, the answers received risk being an extended confirmation of what one already knows or believes in. This directly poses a concern: the risk of confirmation bias. Further to this, scholars in computer science argue that

“ChatGPT-3 lacked intentions, goals, and the ability to understand cause and effect — all hallmarks of human cognition”[4]

Additionally, when specifically considering Diversity and Inclusion training development, AI cannot match the human capacity for creativity, empathy and understanding. Humans can bring their own personal experiences and insights into the mix based on their learnt and lived experience, which unfolds a complex and intertwined existence, not possible for AI to conceive or even relate to.

Why is lived experience essential?

Lived experience, per see, provides a unique perspective on the world. Lived experience is emerging as an invaluable insight beyond the grasp of AI; some argue that “witnessing and describing the experience is a protected niche (…)Little islands of human-ness amongst the rising tide of automated mediocrity”[5].

Inclusion and Diversity training aims to provide a psychologically safe space where a diversity of lived experiences and views of the world are shared and understood to inform our decisions better and result in better outcomes for all. A successful D&I learning leads to new insights, transforming how we think about the world and act towards each other – inclusion and social justice being a moral drive.

Having a human interaction with expertise on the subject matter, complemented with creative and critical thinking, empathy, and compassion, is something AI can not offer in an emotional and equity-intelligent way.

At Inclusive Employers, we listen to understand the needs and aspirations of our clients. We offer bespoke courses, content, and meritorious experts with diverse lived and learned experiences to deliver an interactive, engaging, genuinely insightful, thought-provoking, yet psychologically safe learning environment – a learning experience worth having!

Take a look at our training options

How can AI help us in the inclusion agenda?

A definite advantage of AI is the speed at which it researches vast amounts of data, generating content faster than humans. Artificial intelligence tools can help identify patterns of discrimination in large datasets, allowing us to understand better the scope and magnitude of the problem or inequities. It can also help create e-learning on basic concepts and subjects, such as equality laws and legislation (UK’s Equality Act 2010).

AI can be used to create more engaging and accessible content, making it easier for people of all backgrounds and abilities to access and understand the information they need. For example, ‘for learners who are deaf or hard of hearing or those who prefer to read rather than listen to content, ChatGPT can provide speech recognition services, transcribing audio and video content, providing captions, and enabling learners to search for specific keywords [6]. Finally, AI can provide personalized recommendations and guidance, helping individuals find the resources to participate fully in society.

Present-day challenges of AI and Inclusion, Diversity and Equity

1. AI systems can inadvertently perpetuate biases and discrimination if not designed and trained appropriately. UNESCO states, “the biases that people carry in their everyday lives can be reflected and even amplified through the development and use of AI systems”. In April 2023 researchers found that “Chat GPT displays toxic behaviour more than expected”,[7] and other researchers noted the difficulty to “define what it means to mitigate known racial, gender and religious bias in a universal manner — since appropriate language use varies across context and cultures”[8].

2. Lack of representation and diversity amongst those who are creating and coding AI perpetuates iniquities and influences what AI is used for, not addressing the most pressing needs of underserved communities. For example, women’s representation in AI development remains low, constituting only 12% of artificial intelligence researchers globally and 6% of professional software developers[9]. The risk of AI being designed only to serve a certain part of the world or part of society is a real possibility.

3. Lack of legislation governing the way AI is used. The rate at which AI progresses surpasses humanity’s capability to accompany it with appropriate legislation. Similar to how legislation is taking time to regulate online bullying and harassment on social platforms.

4. Disinformation at a significant scale. Besides factual errors produced by AI (text generated on the statistical likelihood of being true), current AI can generate convincing text and images. Microsoft’s chief economist warned that ‘AI could do a lot of damage in the hands of spammers with elections and so on’[10].

5. Privacy and use of personal and business data– based on this concern, on 31 March 2023 Italian Government blocked Chat GPT use in the country following concerns on data privacy on the basis of European Union law on GDPR (General Data Protection Regulations)

Key points to consider:

Artificial Intelligence such as Chat GPT mimics human-like interactions but cannot replace human lived experience and the associated human capabilities of creative, emotional and equity intelligence in creating a behavioural-changing D&I training or learning experience.
There is a pressing need to broaden the diversity and representation of who develops and codes artificial intelligence and which values it endorses.
There is a pressing need to ensure AI is designed to be ethical and unbiased and that Conversational AI systems are representative, fair, transparent, and accountable.
The pace of AI development surpasses human capacity to reflect on its implications, moderate our relationship with and use of AI, and endorse robust legislation and principled ethics to govern its use.

The potential of this powerful AI is promising and exciting yet scary when we cannot yet foresee the extent to which its impact can have on our lives, sectors, industries, governments and humanity at large.

AI professionals such as Elon Musk (OpenAI) and Geoffrey Hinton (early pioneer of AI and ChatGPT apps who recently resigned from Google unfolding a debate on the risks of AI de-regulated use), have overtly called for a pause in major AI projects[11].

UNESCO has been playing an influential role in international efforts towards the ethical development of AI and in March 2023 called all governments to implement the global ethical framework (signed by members in 2021) by developing respective strategies and policies[12]. Worth noting are the four core values the ethical framework permeates[13]:

Respect, protection and promotion of human rights, fundamental freedoms and human dignity
Living in peaceful, just and interconnected societies
Ensuring diversity and inclusiveness
Environment and eco-system flourishing

While artificial intelligence serves a purpose, the human experience in the Diversity, Equity and Inclusion environment is one that people resonate with. The human voice and lived experience need amplification in the workplace, and a way to support that is to provide a psychologically safe environment and true inclusion ethos.

Enquire about our workplace training options