A strategic approach to responsible AI in 2025
AI technology matures in a fast pace, more impactful AI use cases make it to production and with the introduction of SLM’s and superchips, we also make steps towards more sustainable AI. Great progress related to the technology itself. This article discusses our approaches to adopt AI technology and how we can progress in consuming AI more responsibly.
Responsible AI, a topic worth a book. For now, I cover just a few aspects to consider for 2025, to influence or contribute to. To clarify, let’s look into three questions:
The answers and suggestions may inspire you to reflect on your current business approaches to AI consumption, some of the suggestions are equally usefull for your own productivity cases.
No idea? Also not after consulting Gemini, ChatGPT or Grok? No worries, I think that nobody knows!
Is it even relevant to know how many AI solutions exist? I guess not, however ……
According to the state of AI in 2024 a report just released by TAAFT, 13.734 AI tools have been pushed into the AI marketplace since end 2022. Yes, you read it correctly 13.734. In the same report we can find that there are 272 AI tools for interactive storytelling and 135 AI tools for interview preparation. The abundance of AI tools that do more or less the same, introduces the luxury problem which one to choose. A more relevant question in this perspective: do we really need that many AI solutions for similar tasks? The answer: Obviously not!
The good news is that market dynamics will resolve this issue of luxury anyway, only AI tools with impact will survive. It is up to the AI producers to consider how to spent resources, time, money and energy to make an impact with AI in 2025.
So what is the point?
Consider the total waste of resources (manpower and energy) spent already on development for similar AI ‘solutions’. Even with the understanding that creating a new AI tool or app is not that labour intensive, imagine the benefits we mobilize if we collectively orchestrate the AI tool production only a bit more than we do today. Also, it would be a more responsible approach to the creation of AI tools.
One of the approaches to make this happen is the AI ecosystem. AI ecosystems enable contribution and collaboration on a certain scope of business interest. This can be horizontally or vertically and applies for enterprises and also small business. This more combined and orchestrated approach to AI makes AI development more effective, efficient and more sustainable. Siemens Industrial metaverse is one of the examples of an AI ecosystem in action.
While I refer to the AI ecosystem as a means for a more efficient and responsible approach to AI, it is also, and may be in the first place, an AI strategy for companies to consider. For smaller companies it can be a (financial) challenge to adopt AI. These companies can leverage their AI environment and establish an AI ecosystem to create economy of scale. For large enterprises, the AI ecosystem can be a new product offering or it can be used to improve customer loyalty. For instance, banks can take lead in AI ecosystems to support smaller companies (SMB) in AI development and in doing so improve customer loyalty.
The overall idea is to collaborate for efficient and responsible use of AI.
Congratulations if you could not answer this question or when your answer was: ‘it depends’.
Identifying the best AI tool is not straight forward. In the first place because of the problem itself. I’m quoting my friend Tom Ormsby : ‘AI won’t solve the problem for you, unless you frame the problem correctly. There’s no shortcut to a solution, and every solution has to start with a well-defined use case’.
If identification of the best AI tool is not straight forward, how to interpret lists that circle on the internet telling us what ‘the best AI tool is for……’ ? are these lists useful?
These ‘AI tools lists’ aim to provide some clarity in the abundance of available AI-tools. In itself that is useful and each lists represents a certain value. In particular if time and effort has been put into benchmarking and user evaluation. Unfortunately there seems to be a collective fail to turn this value into something beneficial for non AI experts. With many distinct and even more individual (AI influencers!) attempts to create insight, overall insight got lost. The sheer number of lists circling on the internet already requires an AI-Agent to unravel and consume contents. And with CO2 footprint indicators missing, lists are de facto not helping consumers to select tools based on responsible use.
There is no point in adopting AI just for the sake of the hype. Many use cases can be resolved using AI, however, they can evenly well or even better be resolved without AI. Also, solutions based on more traditional AI techniques or based on smaller models (SLM’s distilled from LLM’s) may do the job equally well. Different techniques with different impact on CO2 footprint. There is a point in considering all options to resolve a problem and use AI responsibly. This also applies for personal (productivity) use. Each time you consider to use an AI solution, rethink your options for a more responsible choice. May be use that tool based on a smaller model, may be use the AI tool from that company that promotes responsible AI or may be don’t use an AI tool at all.
The overall idea is: Know when to say no to AI.
It does make sense if you did not answer this one! And if your answer reflects some form of disappointment: you are not the only one!
The value question is somewhat tricky to answer. Value can be in terms of ROI, it can be in terms of adoption rate it can be in terms of true impact for the world, society and humanity. Irrespective, the tendency is that we seem to struggle to get the best out of what the technology has to offer.
According to the October 2024 BCG report Where’s the Value in AI? 74% of companies struggle to achieve scale and value from AI. The CDO Responsible AI Benchmark Report 2024 reveals a similar picture. With 88% of companies indicated to have at least a (one) AI tool productized, only 37% is able to scale. The report also reveals that majority of companies rely on commercial AI tools and points out a strategic gap between the urge to implement AI and the ability to use AI effectively and responsibly.
Companies may take a use case by use case approach to decide on AI adoption. Arguments for this approach are to keep things small and make value on short notice. This may initially work, does it work in the long run?
Both surveys confirm that making a business case for an individual usecase is difficult, and an important reason for AI adoption stagnating. Jumping from failing business case to failing business case is clearly not the way forward. To add, using AI for supporting functions in the company only, may be needed, it might not make the difference. Recognizing that AI technology can be transformative, requires an approach that does justice to this potential: an AI strategy. A strategic approach includes the company objectives and reimagination of business using AI. Assessment of the company state of AI readiness (data, knowledge, infra structure etc) is part of an AI strategy and this sets the priorities in AI adoption. To add, an AI strategy introduces accelerators to enable the company to transform and to provide guidance on how to consume AI responsibly on daily basis.
The idea is to think big in order to get the best from AI technology.
info@aimetimpact.nl
+31 6 52 03 32 70
KVK: 96361662
BTW: NL005206141B72
