The hype surrounding artificial intelligence seems much less intense than it was just a year ago. The reason is simple: we started using AI in our daily work, and often the results didn’t meet our expectations. We see the potential of AI, but we’re also becoming increasingly aware of its limitations and the challenge of effectively integrating this technology into everyday business processes and work.
This aligns with the Gartner Hype Cycle model, where after a wave of enthusiasm, there’s a phase of reality check and disappointment:
It’s Natural for Things Not to Go Perfectly Now
Examples of disappointments? Here you go:
- The AI Overview feature introduced by Google in May 2024 provided imprecise and often useless answers, especially for unusual queries; as a result, the company reduced its visibility.
- Memes mocked AI-generated responses, such as suggesting using glue as an egg substitute in baking.
- The image generator in Gemini produced images of Asians as Greek warriors or a female pope, while Leonardo.ai often depicted engineers as only Indians (not to mention the notorious twin issue).
- ChatGPT 4, when based on attached data from the Polish Central Statistical Office and without advanced prompting techniques, often fails to list the 10 largest cities in Poland correctly. The likelihood of hallucinations remains high, especially when asking for precise and less popular information.
I suppose that each of us could add our own examples to this list as well.
Adding to this are regulatory uncertainties, copyright concerns, the complexity of corporate processes, and mismatched procedures, which sometimes discourage even trying… A slightly different, broader, and bolder perspective on company data is necessary. The costs of implementing AI solutions and hiring specialists are perceived as high. Trust in AI-generated responses is rather low (rightly so!), which in turn makes it difficult to decide to give algorithms significant control over company processes.
For some companies, emphasizing that they are moving away from AI has become a strongly articulated distinguishing feature. For example, I wrote about the “human only” trend in an article about deepfakes. I consider it a short-term tactic, not a long-term trend, but it also confirms a broader perspective.
Think Long-Term About AI
At the same time, AI is a market segment into which an unprecedented amount of money has been invested, and AI solutions are improving every day. However, this evolution is happening somewhat in the shadow of other events, so the awareness of progress is not as great. This determines the general mood around AI.
It’s worth noting that to harness AI’s potential, competencies are needed. To get any response from a large language model (LLM), you just need to enter any prompt. To get the best possible quality response, knowledge of prompting techniques is crucial.
The same goes for implementing AI tools. Here, too, analysis and sensible calculation of benefits are required. Wisely implemented AI-based solutions will present a huge opportunity for companies; an approach based on exaggerated promises and buzzwords can only lead to disappointment.
Transforming AI Missteps into Milestones for Progress
This period, with somewhat less hype around AI and some companies and their employees emphasizing the limitations of AI over its benefits, is, in my opinion, the best time to act. It’s time to train employees and look for areas where we can benefit from AI’s potential right now; to gradually change our thinking about AI and modernize companies.
AI can be an opportunity not only for companies but also for Poland as a whole (and your country too if you are from outside of Poland). I believe it will be very difficult for us to develop a large language model competitive with solutions from OpenAI or Anthropic. But we have proven many times that we are very good as a nation at seizing opportunities and tapping into existing trends. Over time, I hope we will develop important foundational technologies in the AI field, but for now, we can take advantage of the achievements of big tech solutions, occupy niches, and build competencies.
I would very much like small and medium-sized enterprises to also adopt AI solutions and strengthen their competitiveness. Of course, building competencies is time-consuming and can be expensive, tools have their price tag, and setbacks are inevitable. However, it’s a cost worth paying, as there’s no indication that AI development will stop, and at some point, it may be very difficult to jump on the bandwagon.
Please, don’t stop reading after this sentence: I have set myself the goal of building AI competencies among employees of Polish companies through training (you’re welcome to join). For management, I offer consulting services. At the same time, within my company Oxido, I want to test the AI First approach in some processes. Yes, sometimes progress is slow, but I want to prove that it is possible. I hope to encourage other entrepreneurs to act boldly through successes.
“Sing as if no one is listening.” Let me paraphrase these words – implement innovations and build competencies when others think it’s not the right time yet. Because when the time comes, you will be in a completely different place than the rest. You don’t have to (and shouldn’t) bet everything on AI right away.
There were too high expectations for AI in the short term; sometimes due to the creators themselves who wanted to make a good business or get additional funding from investors. In the long term, artificial intelligence is underestimated.
I invite you to subscribe to the newsletter to receive notifications about future articles and possibly interesting bonuses soon!