30/08/2023
Back to all articles
Everybody talks about ChatGPT, the generative AI system, and everybody tries to figure out how to use or adapt it for the proper company.
ChatGPT is supposed to give access to knowledge in a fast and efficient way. It can help streamline processes and automate in an efficient way.
Many companies – with the idea of being safe in regards to their data and in order to adapt to their specific needs – are working on the development of a proper Chat program.
McKinsey, among the first, has developed Lilli. Lilli helps employees find and get access to case studies, meeting transcripts, and presentations of the company. Via voice command their employees quickly find the document they are looking for.
Bosch founded the ‘Bosch Center for Artificial Intelligence’, a center for AI excellence within Bosch Research. They drive AI projects from the first idea to the implementation, from fundamental research to real-world products. Bosch wants to take AI to the next level making people’s lives easier, safer, and more comfortable. To realize this they work with cross-functional teams leveraging big data from more than 230 Bosch plants worldwide. The goal is to use AI in smart, connected, and autonomous technologies across all business sectors. Bosch collaborates with thought leaders from industry and academia in regard. The research topics embrace deep learning, NLP, Neuro Symbolic AI, probabilistic modeling, reinforcement learning, control, and optimization.
DM, the German pharmacy chain announced this August that they rolled out an AI-based Chatbot, dmGPT (dm Generative Pre-trained Transformer) for employees. They use the same ChatGPT technology in the background but operate on the dm cloud infrastructure. The dm AI can edit texts, support with programming, correct program errors, create concepts, help with research, and create social media posts.
In the Journal of Innovation Management (JIM 8, 1 2020 page 39 – 50 ) 6 levels of Managing AI in a company (AI maturity assessment) are described:
1. Isolated Ignorance – level 0
2. Initial Internet – level 1
3. Independent Initiative – level 2
4. Interactive Implementation – level 3
5. Interdependent Innovation – level 4
6. Integrated Intelligence – level 5
AI may be defined as a “system’s ability to interpret external data correctly, to learn from such data, and to use those learnings to achieve specific goals and tasks through flexible adaption” (Kaplan&Haenlein, 2019, p.15).
Some studies and companies may focus on technological opportunities and solutions, whereas others primarily focus on market applications and use cases of AI. As technology and markets evolve over time AI is dynamic and the scope changes with these developments.
Managing AI in a company happens on different levels:
1. Strategy level
2. Implementation level
3. Operational level
4. Governance and Ethics level
5. Skill Development level
6. Collaboration level
7. Continuous Improvement level
On the first level, the scope is to find a strategy as to where and how to use AI. The key business areas for AI are identified. AI will be aligned with business goals, potential use cases defined and a roadmap for AI integration will be set.
The challenge is to understand how AI fits into the existing workflow, and how it can be improved. Stakeholders have to be involved and respond positively. This is part of level 5 also.
The second level, implementation is when AI systems are developed and deployed, including data collection, preprocessing, model training, and integration with already existing systems.
The challenge within the implementation is to acquire high-quality data, ensure privacy and security, select appropriate algorithms, and avoid biases. Security and privacy are discussed about a lot and claimers try to provide a certain privacy. AI learns by itself, this is the idea and it is used to improve efficiency, productivity, a clear business model serving the output, and the gain of a company. How can security and privacy be guaranteed? This claim seems against the whole idea of AI and is impossible to sustain.
On the third level, the operational level, the day-to-day management of AI systems takes place. This includes monitoring performance, handling system failures, ensuring data quality and security as well as managing scalability, and addressing unexpected behavior of AI models. As for some of these challenges, companies tend to develop their own AI system, especially to enhance security and privacy, which is questionable.
As for Governance and Ethics, establishing guidelines for AI, data privacy and regulatory compliance is a must to responsibly use AI. As companies are still dealing with the steps before and are not really aware of all the possibilities of AI and challenges with respect to ethics, this step may have been left behind so far. However, Ethics and the general normative approach to AI should be thought about from the very beginning, in general, and parallel to the evolution of AI. Defining AI principles, navigating regulations as the EU AI ACT ( http://europarl.europa.eu ai ac, http://digital-strategy.ec.europa.eu), and deciding proactively and in a normative way what we want, what we do not want, and what we have to accept without being able to influence. AI is much more than recognizing speech, writing texts, and streamlining texts, it is an intelligent system that is able to learn by itself and it is a system developed by humans, humans who program it, and humans who sell it.
Level five deals with training and upskilling for employees to understand properly and work effectively with AI technologies. This is when companies are fully prepared and ready to switch to the concrete application of AI systems. The challenge is to identify skill gaps and overcome resistance to change. Skill gaps in humans with respect to AI may lead to something not wanted.
On level six, the Collaboration level, companies do work with the developed AI systems and apply them cross-functional between departments to maximize impact on the various business aspects. The challenge here is to break down silos between teams, empowering communication and transparency, knowledge sharing, and aligning according to single and common goals.
The Continuous Improvement Level is AI intrinsic in a way, it is part of the basic idea. This level is about refining AI models (finetuning) and processes to improve accuracy, efficiency, and overall value over time. A challenge may be to update without disrupting operations and adapting to business needs.
Gartner introduces a compact five-level AI Maturity Level:
Level 1: Awareness, which is early AI interest with the risk of overhyping
Level 2: Active, AI experimentation, mostly in data science concept
Level 3: Operational, AI in production, creating value by process optimization for product/service innovations
Level 4: Systemic, AI is pervasively used for digital process and chain transformation, and disruptive new digital business models
Level 5: AI is part of business DNA
(source: gartner.com/SmarterWithGartner, 2019)
The level of AI maturity assessment may vary in smaller and middle companies. Some may be in the early stages of exploring AI possibilities while others may have implemented basic AI applications.
Assessment of AI maturity involves evaluating factors like data readiness, technology infrastructure, skill sets, and alignment with business goals.
The process seems rather inductive, scanning superficially AI applications and implementing them just to jump on the train, to be part of the hype.
Instead, apart from the whole ethical side that is clearly underestimated and scarcely treated, we all should analyze options and possible fields of applications, evaluate cost and outcome ( do we have measurable KPIs?) and proactively decide when where, and how to integrate and implement AI in our businesses with what result and consequences. Process in a deductive way.
A hype easily becomes inflationary. Instead, humans should hold up this wave and ride it in an intelligent and sustainable way, getting the most out of it according to their and the companies’ goals and purpose.
The intelligent human is the decision maker, not the intelligent machine!