From co-pilot to co-worker: GenAI
Generative AI has taken the industry by storm. Fatih Ogun, Head of Strategy at Akbank, tells us about the points we need to consider when it comes to AI stepping in to become a co-worker.
Generative AI has taken the industry by storm. Initially, the tools were used primarily for academic purposes, yet they have become near-commercial grade applications with features like text generation, summarization, question answering, conversation, and image generation. The accuracy, fluency and articulacy of the outputs are surprisingly good most of the time and in many cases it is hard to distinguish AI-generated content from human-generated content. Achieving this level of quality also raises the possibility that these features could be applied to business settings. This naturally brings us to the question: “Can we use these algorithms as co-pilots or maybe even co-workers?”
Creating such ‘automata’ structures is an age-old dream for humans. As you are doing your daily work or completing a specific assignment, receiving high-quality, on-demand help is priceless. On the other hand, the help you are getting should be accurate, concise, unambiguous, and should correctly address the challenge at hand. For instance, in software development, getting help for repetitive code is applicable and timesaving, but providing to-the-point support for complex algorithm or interface development requires greater precision and skill. In such cases, the concept of co-piloting seems to be working and apparently its application radius will widen in the near future. However, when it comes to AI stepping in to become a co-worker, there are some points we need to consider.
• Subject matter knowledge. Having a co-pilot that can work on a variety of subjects is great, but for a co-worker, specific subject matter knowledge is required for achieving certain tasks (such as delivering a project output). Existing programs are able to sift through different knowledge repertoires in addition to their base language models, but subject matter expertise requires more dedicated training and an understanding of historical projects. In other words, depth of knowledge gains precedence over breadth of knowledge.
• Autonomous work capability. The co-pilot, in a basic sense, relies on prompts. You ask for something or give a command and the co-pilot model produces an output. As a co-worker, the model should list, plan, and work on tasks autonomously. In other words, it can take the initiative. Achieving this is not just dependent on establishing or training a language model, it also requires careful consideration of the relevant context and stakeholders. In addition to the subject matter, the model has to evaluate these project-specific factors.
• Deep collaboration. By this term, we mean that GenAI should go above and beyond prompts and managing conversation discourse. The model must keep the context of the work in mind, the objectives of the study, and what the human co-worker is expecting and aiming to achieve. Just like in deep learning, a multi-layered approach is required. Without collaboration at depth, the model, no matter how good it is, carries the risk of becoming a reactive device working on prompts.
• Use case deployment. As in the software development discipline, managing the deployment order of new features is critical to the success of a whole system. The same applies for achieving co-worker level capability and reliability. It is important to select the use cases carefully and it is crucial to make sure that you do not jump to the next level before comprehensively achieving the objectives at the current level.
As we said in the beginning, this is an age-old dream. Achieving it in the right way could clear the way for considerable productivity gains and elevate our daily work structures to the next level.
The views expressed in this article are the views of the author only. This article provides general information and a point of view; it should not be considered as professional advice.
Digital Reinvention community
With Qorus memberships, you gain access to exclusive innovation best practices and tailored matchmaking opportunities with executives who share your challenges.