Issue #3: Leveraging LLMs - A New Frontier for the Managerial Class
Maintaining Your Human Edge as Machines Get Smarter
The World Economic Forum (WEF) recently released an extensive study titled “Jobs of Tomorrow: Large Language Models and Jobs” analysing how innovations like ChatGPT and LLMs could transform and disrupt the workforce.
The findings are concerning to say the least - up to 70% of management analysts' work duties could be automated by LLMs.
Some specialists like financial managers may see over 75% of their responsibilities exposed to takeover by LLMs.
Across functions, IT (Information Technology) and finance roles are highly susceptible with over 70% of tasks at risk.
Even roles in operations, marketing, HR, and legal could witness 30-60% of responsibilities automated.
Clearly, wide swaths of the management profession face disruption as AI’s take on analytical, coordinating, and data-driven work.
Repetitive tasks like processing information, preparing reports, and monitoring workflows are primed for automation.
However, while algorithms can replicate discrete tasks, the report stresses that the essence of leadership remains innately human.
Skills like creativity, imagination, empathy, judgment, communication, and strategic vision cannot be replicated by AI. These distinctly human capabilities will only increase in value as automation advances.
If there is one positive message from the report it is that “Adopting generative AI, with its potential to increase individual worker productivity, could fuel significant job growth.”
The question that stood out for me was - From a managerial standpoint, what insights or lessons can we glean from this report?
The stakes are clearly high for managerial class who oversee workforce strategy and planning for entire teams or functions.
Adapting to AI-driven change in workflows and job architectures may prove more challenging for leaders than individual staff members.
In this newsletter, we will dive deeper into:
Why we should be looking at a broader impact of LLMs in the workplace that go beyond text and generation, looking at multimodal LLM capabilities.
The specific functions and responsibilities most exposed for middle managers and top executives.
Practical steps managers can take today to integrate AI-powered work while amplifying their human strengths.
Proactive managers have an imperative and an opportunity to shape how AI integrates into their organisations.
Managerial insights and actions today will determine whether these ‘thinking machines‘ elevate or impair the workforces they manage.
Don’t Ignore Multimodal LLMs
While the “Jobs of Tomorrow” report provides useful initial analysis, its scope is limited to large language models (LLMs) that process text alone.
However, AI capabilities are rapidly expanding into processing images, speech, video and more.
A key point that few recognise is that Multimodal LLMs belong to a distinct category of models, setting them apart from single-domain AI models focused solely on computer vision, speech, or video.
For example, computer vision focuses specifically on processing visual data like images and video.
In contrast, multimodal AI integrates information across text, images, and potentially audio.
This enables more interactive conversations about visual content, richer contextual understanding, and inherent accessibility through generating text descriptions.
A key advantage of multimodal LLMs is their versatility - they can be quickly deployed across diverse domains without requiring training from scratch on niche datasets.
For instance, GPT4V can already identify plants, animals, products, places, celebrities, and more from an image without specialised training.
GPT-4V represents a major advance in multimodal AI through its ability to fluidly combine image and text processing.
It can answer questions about images, identify and focus on elements within complex scenes, interpret medical scans, perform common sense reasoning about visual depictions, transcribe and translate visual text, categorise images when given examples, interact with user interfaces and physical spaces, assess aesthetics and defects, and leverage news retrieval to provide context about images and much more. This 166-page paper is worth diving into to get a sense of how powerful these multimodal LLMs are.
Now, lets apply the multimodal LLM lens to the report.
One insight is that while computer vision lays the groundwork for visual data interpretation, multimodal LLMs creates a more versatile, creative and integrated human-machine experience.
This fundamentally expands the impact horizon for physical occupations previously considered safe from disruption.
For example:
In healthcare, multimodal AI can analyse medical scans and discuss the results conversationally. Studies show it can outperform radiologists in diagnosis.123
For supply chain roles, multimodal AI can combine computer vision with databases to coordinate end-to-end warehouse workflows.
In agriculture, it can optimizs planting and harvesting by fusing computer vision, crop patterns and weather data.
For manufacturing, AI assistants can provide visual and verbal instructions to guide workers.
In education, combining voice recognition, sentiment analysis and language processing enables next-generation tutoring tools.
For media roles, multimodal AI can generate video and audio content, not just text.
In other words, multimodal AI could significantly expand exposure levels for many jobs beyond the 70% suggested for management analysts in the report.
Importantly, the expanded capabilities could increase automation potential for certain tasks, rather than only increasing augmentation.
For example, computer vision could automate visual inspection in quality control or warehouse operations. Sentiment analysis from customer calls could eliminate the need for manual call analysis. AI-guided industrial robots could take over assembly line tasks.
While the text analysis focuses on augmentation of analytical roles, multimodal AI unlocks automation potential for many physical job categories as well.
This underscores the need for holistic workforce planning and skills development to navigate the transition.
In addition, the report's near-term, US-centric, text-based focus severely underestimates the scale of global workforce disruption likely in the next 5-10 years.
Rapid technological progress means managers cannot simply extrapolate from today's limited AI capabilities when planning for the future of work.
As computational power continues growing exponentially, increasingly multimodal AI will penetrate roles we consider “safe” today.
Implications for Middle Managers
Middle managers overseeing analytical functions like finance, IT, and operations are likely to see the most immediate impact as AI adoption grows.
As repetitive data tasks get automated away, managers' day-to-day responsibilities will transform.
For example, monitoring financial performance via written reports and spreadsheets could be handled by AI synthesising data across formats like speech, imagery and text.
Finance managers would then focus more on strategic analysis, long-term planning, and high-value client interactions.
Similarly, multimodal AI could help operations managers coordinate across teams using a mix of visual data from sensors, verbal communications and database access.
This provides space to address more complex challenges needing human judgement.
The key is that automation should aim to offload repetitive work and empower people-focused priorities - not wholesale replace jobs.
Managers adept at communication, strategic thinking and nurturing talent will remain irreplaceable. But those failing to reinvent workflows risk obsolescence.
Besides core analytical functions, quality management and customer service present more opportunities as AI incorporates computer vision, speech recognition and natural language processing.
For instance, production line defects could be automatically detected and logged in natural language. Or customer calls could be transcribed and analysed to find pain points.
To adapt, middle managers should conduct audits to identify the most automatable tasks within their team’s workflows using a multimodal lens.
Implications for Top Managers
While middle managers handle day-to-day operations, top executives focus on high-level strategic planning and decision-making.
Although their responsibilities are less routine, multimodal AI offers significant potential to enhance and accelerate executives’ work.
Multimodal AI can enhance tasks like reporting, analysis, and forecasting for top executives.
Here are some specific ways:
For reporting, multimodal AI could monitor customer sentiment on social media through text and video, competitor press conferences via voice transcripts, and supply chain disruptions through visual data from satellites and sensors. This provides a holistic real-time view.
For strategy analysis, multimodal AI could process past sales data, customer interviews, support calls, marketing campaign images/videos, and market research reports to identify new pricing models, partnerships, and product innovations.
For forecasting, multimodal AI could synthesise earnings calls, industry conference presentations, demographic visualisations, clinical trial results, and policy proposals to model demand for new products and services.
Multimodal AI could also create interactive visualisations and dynamic dashboards bringing together diverse data sources to highlight insights and trends for strategy discussions.
The key is establishing human-AI partnerships that combine machine scale and speed with human judgment and imagination.
While purely creative tasks remain largely human, multimodal AI will increasingly generate new possibilities.
Top managers should pilot these AI collaborations on high-potential initiatives where fresh strategic thinking is needed.
They also must double down on capabilities like inspirational communication, empathy and ethics to provide vision and oversight.
The machines will crunch data - but human leaders must direct how it’s applied.
Takeaways
As this newsletter outlines, managers must closely track the rapid evolution of AI across all modalities and proactively adapt workflows and workforce plans.
Here are key takeaways for preparing your organisation:
Teaming, Not Automation: Institute human-AI collaboration models rather than simply automating existing processes. Blend machine and human strengths.
Re-skill and Recruit for the AI-powered Workforce: Re-skill employees by emphasising creative, interpersonal and strategic skills. Recruit those with capabilities to interpret and manage AI.
Continuously Evaluate AI Impact: Audit workflows to identify automatable tasks. Continuously monitor emerging AI capabilities.
Accelerate Automation of Repetitive Tasks: Use AI to automate repetitive analytical and operations work. Empower managers towards more strategic priorities.
Build an Agile, Responsible AI Culture: Foster adaptation while implementing ethical AI guidelines to address biases and build trust.
The managers who proactively reshape workflows, amplify human talent, and integrate AI's capabilities will thrive.
This new era requires reinventing processes, not just plugging in technology.
With strategic preparation, AI can help unlock human potential and uplift management to new heights.
🚀 Upcoming Free 1 Hour Webinar: Elevate Your Professional Game with AI!
Are you a forward-thinking manager seeking to get ahead of the AI disruption? Join my upcoming webinar for an inside look at how to prepare you and your organisation to develop the human+AI teams of the future.
In this interactive session, we’ll explore cutting-edge generative AI capabilities like text-based and vision-based models and a live lab where we’ll experiment with real-world AI use cases you can implement.
Reserve your spot now for this complimentary session - space is limited.
You'll leave inspired, knowledgeable and ready to lead alongside the machines!
⏰ Reserve your spot! ⏰