5.4 Ready-Made-AI Adjusted
In this project, Ready-Made-AI was initially used to narrow down exactly what Bluefragments were expected to develop for Bestseller. Hjørnholm’s team started with three potential projects, but through a discussion facilitated by the Ready-Made-AI workshop, the parties narrowed the project down to the internal dispatching system (Appendix 3, l. 349-355).
The depth of the discussion regarding AI were perceived somewhat differently by the two parties. Martinsen believed that a broad understanding of AI technology was evident from the beginning:
“They had the overall perspective. But we explained some of the core.” (Appendix 3, l. 414
Clearly, Martinsen did not feel that Bluefragments had to explain quite as much as they might have done with other clients. Nonetheless, Hjørnholm still felt that Bluefragments needed to get the Bestseller team up to speed on the basics of AI:
“Well, first of all being very low practical, which was also one of the things that I emphasised to Thomas, that were needed - as we are not rocket scientists or anything in that regard. We really needed to take that approach, you know, keep it simple. Try not to jump into the pool at the deep end.” (Appendix 6, l. 198-200).
Martinsen felt that Bestseller already had a high level of knowledge about the subject because they were able to deep dive into the technology quite early on in the project (Appendix 3, l. 365-367).
Hjørnholm looks at it differently because Bestseller had a high working standard. Meaning that they
had more technically insightful questions, as they had a high degree of initial knowledge. Bestseller’s
technical appreciation meant that they needed an advanced understanding of the technology to be
comfortable using AI. We believe that Ready-Made-AI has been used to discuss what was technically possible for Bestseller to implement (Appendix 3, l. 349-361), rather than as a theoretical discussion about what AI is. However, we believe there has still been a strong focus on the potential of AI technology and the aspirations both parties had in that regard. Meaning, that the discussion has not been limited to what was technically possible for Bestseller.
Since Ready-Made-AI was used to frame a discussion it was therefore mentioned during the collaboration, but Martinsen was unsure whether Hjørnholm understood that this concept was being used. In fact, he was not sure that many, if any, of his clients think about the terminology during a project:
Again, we talked a lot about it ((Ready-Made-AI)), initially. But I think many customers they actually don't think that much about terminology. Once we are in the project. They think more about their business and how it fits in.” (Appendix 3, l. 359-361).
Martinsen’s comment would seem to suggest that Ready-Made-AI has never been a forum for a conceptual discussion about various definitions of AI. Nonetheless, facilitating a conversation remains important to Martinsen, as he sees it as Bluefragments responsibility to create a common language with the client about AI. Unsurprisingly, Martinsen also remained convinced that it was necessary to create a shared terminology with Bestseller about AI (Appendix 3, l. 371-375).
Martinsen’s focus on creating a shared terminology has had an effect on Bestseller, as Hjørnholm acknowledges:
“Bluefragments taught us about the importance of clean, structured data. Next AI endeavour will most likely reflect that in our approach to data capture and hygiene.” (Appendix 6, l. 805-806).
Here is a clear example of how Bluefragments have managed to create an understanding about how data should be treated and addressed when developing AI - specifically for this case. However, it remains clear that the establishment of a terminology is focused on creating a shared appreciation about the technical aspects of developing an AI, and not about the conceptual differences in Martinsen’s and Hjørnholm’s definitions of AI.
Nonetheless, a conceptual discussion might have been necessary as Martinsen believes that
he and Hjørnholm shares the same perspective on AI, when we asked whether he thought they
shared the same understanding he answered:
“Yes. I think Lars ((Hjørnholm)) is a person that is extremely curious about technology and is super bright and understands the possibilities about technology.” (Appendix 3, l. 365-366).
Martinsen clearly endorses that there was a shared understanding, not necessarily in their
definitions of AI, but in the vision they have for the technology’s use. The reason for that answer is based on Hjørnholm’s technical appreciation of technology itself and the possibilities found in the use of these technologies. Hjørnholm also gave the impression that he and Martinsen shared the same perspective on the project:
“It never stood out to me, at least. As, no I think the perceptions were (.) kind of similar to a large extent.”
(Appendix 6, l. 218-219).
As was evident in our findings, they do not agree on what the definition AI is. Hjørnholm believes that genuine AI is likely unattainable, therefore it could be argued that Bluefragments might never be able to provide something, which actually lives up to Hjørnholm’s definition of AI. Nonetheless, this argument has proven false, as Bluefragments have still been able to provide a solution, which has lived up to Bestseller’s expectations. Despite the fact that there was no agreement on the definition of AI, Martinsen believes the project was a success:
“Definitely a success and the fact that we did that in three days was almost unbelievable.” (Appendix 3, l. 383-384).
Bestseller share the same positive the perspective on their work with Bluefragments:
“Professional yet down-to-earth approach to doing business. They are pragmatic and portray a sense of sticking to proven methods, which impose a feeling of quality and calmness. There is a certain “feel-good” vibe to their dealings. It made us feel comfortable and sparked a creative productiveness that we didn’t foresee.”
(Appendix 6, l. 808-810).