14 March 2024 marks the anniversary of ChatGPT 4, undeniably one of the decade’s pivotal technological breakthroughs, on par with the invention of the internet or microprocessors. The introduction of this chatbot, adept at seamlessly answering questions on virtually any subject, sparked a global frenzy, making it the quickest app to amass over 100 million users. 

As a result, the global business community’s embrace of artificial intelligence technology transitioned from a mere trend to a transformative power. The rapid pace left little room for debating AI’s utility, propelling us instead towards leveraging its practical applications for substantial business benefits. Deloitte Insights 2024 highlights this shift, noting that 64% of CEOs report continued pressure from investors, creditors and lenders to fast-track the adoption of generative AI. 

This applies to small and medium-sized enterprises as well, even if they may lack a dedicated technology department. One of the groundbreaking characteristics of the new technology is that setting it up doesn’t require a team of experts. While it’s true that many businesses don’t know how to identify strategic machine-learning opportunities or how to turn them into disrupting products or services, the reality is that thousands of people launch AI engines on their computers every day, weaving the technology into tasks that range from drafting emails and compiling reports to translating texts and crafting presentations. The broadest use of the technology to date has been in content creation processes, as content marketers were among the first to go from just playing around with AI to using AI in their daily work. 

Balancing innovation with the human touch

From the outset, the language services industry has been at the forefront of developing multilingual Gen AI solutions that offer: 

  • Authoring help 
  • Content variation by target audience 
  • Tone inconsistency checks  
  • Content classification 
  • Term extraction  
  • Language quality assessment 

At Sandberg, we’ve embarked on our own artificial intelligence journey. For our clients’ baseline translation needs, our machine translation engines still outperform other language learning model (LLM) options, but the GenAI launch prompted us to investigate AI’s potential for content creation in our marketing team and for our own internal documentation needs. 

The results to date have been nothing short of spectacular, even though every instance of the content generation process has still required substantial human involvement. Humans play a pivotal role in generating new ideas and seeking out high-quality references for content. And they have a crucial role at the end of the process, refining the machine-generated output, rearranging the text for better coherence and adjusting the language to align with the brand. For us, this final step is nothing new, thanks to our many years of experience in machine translation post-editing.

Without the human touch, maintaining Sandberg’s unique tone of voice would not have been possible. This is what brings us to our core statement: in the era of artificial intelligence-powered machines, the human element is vital to crafting content, products or services that enrich and redefine human experiences. When almost everything we know will be sourced by our inorganic colleagues, the human element stands as a key value differentiator for brands aiming to connect with their stakeholders at a deeper level. 

Generative human creativity

Let’s be honest: the cost of human touch and authenticity is prohibitive for many companies. The time-saving benefits offered by AI are undeniable. Yet, the challenge isn’t in using artificial intelligence; it’s in identifying each use case and determining the most effective workflows. The indiscriminate application of artificial intelligence in content creation is precisely why our LinkedIn feeds are now saturated with posts that mirror each other. This type of content lacks the unique ability to differentiate or spark interest in a brand.

The capabilities of tools such as ChatGPT, Hey Gen or Copilot are crucial in enhancing the efficiency of documentation, technical writing, legal and marketing teams. But how does a company ensure consistency, maintain specific terminology and preserve the tone of voice in its content when different teams employ different tools? And what happens when the content spans multiple languages? It’s fundamental that outputs such as corporate reports, marketing campaigns, organisational policies or legal documentation undergo human scrutiny and editing. This human-in-the-loop step upholds the brand’s consistency and ensures accuracy and adherence to ethical guidelines. This is where Sandberg comes in to offer professional help to businesses that are turning to artificial intelligence to enhance their content creation processes.

Solutions for content creators using AI

The solution we typically recommend to our clients starts with a process where they leverage an AI engine to generate content. Our team then refines the output, employing terminology databases, style guides and any documents containing instructions that offer insights into the target audience, market or the specific stage of the customer journey associated with the content. 

This is a process we’ve seen successfully implemented at numerous companies. To build a seamless workflow, we organise an initial meeting between our team and the client’s teams. It’s crucial to discuss the technologies involved, to understand the brand identity and to address all the project management aspects, including reference documents and timelines. 

The process can be applied to any monolingual content the client produces with the help of AI, and the output can then also be translated into further languages. 

Other solutions for companies using AI

Although marketers were among the first to adopt AI, the technology’s applications extend well beyond content creation. Many companies are now leveraging AI to process data more efficiently. 

One scenario where Sandberg offers customised support involves data annotation solutions, especially when the data is in languages not native to the client’s organisation. Our team manages the data tagging process to ensure that the client’s machine learning model can effectively recognise the data. 

Other scenarios in the training of natural language processing models where our expertise has emerged as a critical helping hand for businesses include training AI-powered devices. Such training may require human-produced or human-checked content like voice recordings, video collections, text input or prompt design.  

Amongst the many challenges around artificial intelligence, the most serious one is that it’s perceived as a one-push button – a technology that promises to deliver correct answers reliably while hiding the process that produces them. Or just a swift path to cost reduction via automation. 

To avoid the pitfalls of this approach, it’s essential to understand that value extends beyond the end product and encompasses the process itself. A “black box” system offers limited control capabilities that significantly hinder essential aspects such as traceability, creativity and ethical judgement. The best verifiable results are achieved when AI is used to augment, rather than replace, the skills and expertise of humans.

Artificial intelligence, Content creation, Newsletter March 2024