Published in


It is not the individual that will bring the most value in AI.

Why Organizations Fail at AI Development.

Relying on small teams or individuals is not enough to build solutions for real-world adoption. Instead, organizations need to harness crowd wisdom.

(This article is co-authored with Yang Gao and Michael Burkhardt)

A century ago, in 1906, statistician Francis Galton showed the power of the crowd. At the International Exhibition in London, he asked the people to guess the weight of an Ox and the average of what the crowd guessed was 1,197 pounds. The actual weight was 1,198 pounds. Meaning, the crowd´s collective judgment was almost perfect.

Are we smarter together?

The democratization of AI knowledge

Wisdom of the crowd is a known phenomenon, still, companies focus on selecting and working with people at the top who are highly specialized. This happens under the assumption that they have better knowledge, which is partially true, as knowledge was not easily accessible to all.

However, with online education, knowledge, especially in the field of AI, is now accessible to all. One does not have to go to a university to learn the same concepts as someone at Stanford or MIT. Apart from access to education, building real-world AI solutions comes with several other challenges such as rapid advancements, the need for diverse skillsets and perspectives, data challenges, etc., all of which cannot be solved by a small group of people in an efficient manner.

In order to access this distributed knowledge as an organization, the approach to building AI needs to be updated.

How to harness the power of crowd wisdom

Rather than focusing on a few individuals, at Omdena we foster a collaborative and community-driven approach that brings in people from different backgrounds and various skillsets to work together in the most effective way.

Not everyone needs to be an expert in the field but all have the motivation to learn and contribute. The environment is self-organized with no top-down management but instead bottom-up driven development to stimulate innovation and creativity.

The problem to be solved is broken down into smaller tasks by the community through a collective decision process and the individual tasks are picked up by several task groups.

What seems chaos at first, one starts to appreciate the experience and realize the relevance of working in self-organizing environment, where crucial conclusions and optimal solutions eventually emerge as winners. It is very well welcomed and valuable experience. It makes me proud to be a part of it.

— Vjeko Hofman, 10 years of experience as a Software Engineer.

Depicting the division of work in a collaborative model

The benefits of collaboration for organizations

1. Overcoming data challenges

Closing data silos

In many organizations, individual teams don’t have access to all the data because of cultural reasons or a lack of integrated systems (e.g. Wealth and Retail banking don’t share the same systems). However, one of the central mantras of machine learning is “more data is better”, and a collaborative environment will stimulate data sharing right from the start breaking old-fashioned structures.

Generating high-quality data

As said before, in the world of machine learning, data is the key. It is not a sophisticated algorithm or a better team, but it’s the team with the better (and more) data that wins. A collaborative community behind a problem will find creative ways to identify additional data sources as well as to generate more data in a way faster way.

In our collaborative approach data tagging is the fuel for our models. Some task groups work on preparing the data while others are responsible for identifying and then trying out various machine learning models simultaneously to find the best-fit model.

2. Building more ethical systems for real-world adoption

One of the biggest difficulties for AI or ML products is a lack of trust. Millions of dollars have been spent on prototyping but with very little success in real-world launches. Essentially, one of the most fundamental values of doing business and providing value to customers is trust, and Artificial Intelligence is the most-heavily debated technology when it comes to ethical concerns and related trust issues.

Trust comes from involving different opinions and parties in the entire development phase, which is usually not done in the prototype phase.

Many failed AI applications have shown us where fast and intransparent development can lead to. Examples cited include image recognition services making offensive classifications of minorities, chatbots adopting hate speech, and Amazon technology failing to recognize users with darker skin colors.

To solve these problems, the community serves as a validation testbed before putting the solution out into the world

3. Reusing code and knowledge

Photo by Riccardo Annandale on Unsplash

While traditional AI competitions promote doing the same work multiple times by different people, collaboration promotes the optimal distribution of work. AI talent is an expensive resource that should not be wasted. Reusable modules for the whole organization will save a lot of time and effort. Having an open collaborative dialogue will allow more extensible and reusable AI modules.

4. Integrating with an older system

This is another reason why most AI systems fail. Traditional organizations have numerous systems, processes and thus numerous failure points. Questions like, ‘How to create awareness for the solution?’, or ‘How to integrate with older systems?’ need to be answered.

Collaboration between different teams is key to ensure nothing breaks when a new solution is implemented. Teams can also leverage each other’s work to further their AI transformation, which can only occur if diverse stakeholders are involved early in the process.

5. Combing various models to increase performance

A common practice is to pitch one model’s performance against another. However, ensemble models, in other words, combining diverse model types often performs better. In a collaborative environment, this is a natural practice. and through collaboration, we can try five to ten different models and combine the best ones. In our latest challenge to prevent Sexual Harassment, we combined various model which gave us a breakthrough in performance.

6. Empowering people

Lastly, let us talk about the huge social impact that collaborative AI development creates. Traditional organizations, still struggle with both attracting top AI talent and having the expertise to hire the right ones. Through collaboration, people in other departments, who may have the fear of losing their job but are willing to learn new skills, can be retrained by building a real-world solution.

Collaboration is the future of AI because people around the world are eager to work together, solve meaningful problems, advance their careers, and make an impact not only in their lives but also for others.

Phenyo from South Africa

If you want to be part of the #AIforGood movement, join our global community as an (aspiring) data scientist or as an organization.

If you want to receive updates on our AI Challenges, get expert interviews, and practical tips to boost your AI skills, subscribe to our monthly newsletter.

We are also on LinkedIn, Instagram, Facebook, and Twitter.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store