When embarking on this path, we must take into account that many companies do not include the entire value chain, so we must identify the areas in which we can act directly and those that must be developed in collaboration with others.
From the point of view of companies, it is about turning concerns into challenges.
Three steps support the process applied in open innovation:
- Definition of challenges.
- Search for solutions.
- Fit the most suitable answer.
Business Models Based On Ai
Although we may have the impression that Artificial Intelligence has superseded software, AI has become an enhancer of software.
At this moment, we can group the software into two large groups:
- Software 1.0.
- Software 2.0
Software 1.0 uses logic and data as input elements. It applies programming and provides a result that is our goal.
Software 2.0 is carried out in two phases:
- In the first, the objective and a large amount of data are used as input elements. By applying learning models, logic is obtained as output.
- In the second phase, the logic generated in the first phase is used, together with new data, to achieve the objective through inference.
Software 1.0 only allows tasks to be tackled without uncertainty, that is, with precise rules.
Software 2.0 gives us the opportunity to develop tasks that are difficult to describe with precise rules and that move in uncertainty.
Starting from this basis, three types of business models based on AI could currently be described:
- Full Stack. It is a closed proprietary format. An extensive database is created on which the model relies, providing the result in its App.
- Foundation. In a format that includes investments that allow you to expand the set of data on which the model is based, increasing its effectiveness. Other users are then allowed to access the results through an API.
- Frontend. They are formats that take advantage of the Foundation model to access its API, reducing implementation time and investment costs.
One of the most recent cases of the Frontend model is that of LuzIA. The personal assistant that uses the ChatGPT API and is integrated into WhatsApp to make our lives easier.
Benefits Of Open Innovation In Ai
The key is to share, collaborate, learn from others, and share again.
The aim is to eliminate the silos of information and knowledge that currently exist in companies and organizations of all types.
The goal is to implement AI reinforcement learning.
Challenges And Barriers To Implementing Open Innovation In Ai
We can highlight three concepts:
- Fear of being copied.
- Flight of talent due to lack of investment.
- Fear of regulation.
OpenAI Has Publicly Shared Its Model Because:
- He is not afraid.
- It is expensive to implement.
- Only some people know how to do it.
Regarding regulation, Sam Altman (CEO of OpenAI) has stated in a recent interview with the Financial Times that, as the European Union is approaching law regarding OpenAI, it is not ruled out that they will have to stop operating in the territory. Much more challenging and stricter legislation is proposed than what exists in other countries, such as the United States, and Chat-GPT could be in danger on the old continent.
This could lead Europe to an insurmountable loss of competitiveness with other countries, more willing to embrace this new era of AI.
Privacy & Security
Data is the fuel of AI.
It is essential that the tools and data models on which they are built guarantee compliance with data protection regulations.
Business users must be satisfied that they are not breaking the law when using these tools.
One of the most used formats to guarantee this is the anonymization of the data and its elimination once the period necessary for its anonymization has passed.
The investment necessary to launch these projects is low. The problem is that skilled people are needed to train the data models on which AI solutions are based. And in Europe, this type of talent is scarce. If we want to take advantage of this train, quick and forceful measures will have to be implemented to change this situation as soon as possible.