Launching an AI project: first steps
In over 2,000 enterprises surveyed, 47% of these have already launched or are already using AI in their organisations. A lot of businesses are on the way to becoming empowered by AI, however, there are still a lot of businesses out there that struggle to understand AI, its value and what it takes to build.
The problem formulation is the very first issue for your team. In this phase, the teams will have to define the business needs to address. Then they will have to identify the expected outcome, and even more so, what various teams are going to use that outcome for. It’s important to make sure the problem isn’t spurious and it can actually bring real value to the business.
In this blog, we will give you suggestions on things that need to be prepared ahead of launching an AI project.
Getting familiar with AI
Robots and self-driving cars are still the main two things that come to mind when people think of what AI represents. But AI is much more than that. Although it’s easy to think that AI is a mighty technology with the potential of transforming companies, determining the actual role of AI within a company is still not clear. People are still learning about the specific functions that various AI products can perform, or what exactly are the benefits and the issues that they can tackle.
We suggest to start with increasing your understanding of the technology and areas such as Machine Learning and Natural Language Processing. Without a basic knowledge is difficult to figure out what business objectives to pursue, what kind of technology you need and how it will work inside your organisation.
You should take advantage of the wealth of online information and resources available. For those with time to teach themselves there are a wealth of courses online now, here are just a few:
General Assembly: Introduction To Data Science & Machine Learning (moderate technical capability)
Coursera: AI For Everyone (beginner)
Oxford University: Artificial Intelligence Programme (beginner)
Google AI: Learn with Google AI
Alternatively, for those with busy lives, you can always talk to a data expert or go one step further and book an exclusive AI innovation workshop to walk through real applications of AI within your industry and your business.
Prioritise concrete value
Once you’ve gathered an initial knowledge of the technology, start thinking about how you can add AI capabilities to your existing products, services and processes. More importantly, have in mind specific use cases in which AI could solve business problems or provide demonstrable value.
Explore our AI blog section including blogs such as “AI for Insurance”, “4 applications of AI in Marketing”, “AI for Supply Chain”.
External information is helpful, however, you should brainstorm internally or get help from data experts and AI consultancies. Every business is different and can unlock AI value in different ways.
It’s critical for the success of the project that whoever is planning the launch of the project, fully understands what are the major pain points of the business to prioritize.
You should use tools to look at the dimension of the potential and the feasibility and put them in a 2×2 matrix. This will help you prioritize based on short term visibility and know what the financial value is for the company.
Address the internal capability gap
If the discovery phase is quite creative and aspirational, at this point you’ll need to address internal capability gaps that could create an obstacle to the project.
There is a severe difference between what you would like to accomplish and what your internal capability is able to deliver within a given time frame.
By involving different teams as well as third parties, you should be able to assess what your organisation is capable of and what it’s not before launching the project.
The internal capability gap is an issue a lot of businesses are trying to tackle by training teams and hiring new employees to cover missing skills. However, the demand for data science, AI and Machine Learning skills are increasing, forming a gap between the supply and demand. Hiring isn’t simple as many don’t have a clear understanding of the skills required for the project. Not only, data scientists are generally difficult to retain and their training requires time, stopping you from reaching business-impacting results as quickly.
Here is a guide to why scale-ups should outsource their data science. Not every company will want to, but in the early days it will help get a project off the ground much quicker, and there’ll be valuable learning through osmosis.
Prepare your data
Data is the main element of any AI project. Before implementing these technologies into your business, you need to clean your data to avoid a “garbage in, garbage out” scenario.
Data is typically spread out in multiple data silos of different systems and may be in the hands of different teams with different priorities. A very important step toward obtaining high-quality data is to form a cross-business unit taskforce, integrate all different data sets, sort out inconsistencies for accuracy, with all the right dimensions required for Machine Learning.
Here are five main questions about your data to be addressed:
- What data sources do you have available?
- What is the quality of the data (there may be gaps or it could be messy or inaccurate)?
- Do you have enough data (the amount of data required can depend on the model chosen for the given project, so check this with a data scientist in the research stage)?
- How is the data generated and collected (a real-time data environment is very different to deal with than a static database)?
- Do you have permission to use the data as planned (think GDPR)?
With clean and tidy data you are all set up for a successful project. But without you could set yourself up to fail and waste everybody’s time. This is the most fundamental part of AI projects success.
Prepare your employees
With the implementation of AI systems, your teams will be provided with additional insight and automation. They will have a tool to make AI a part of their daily routines rather than something that replaces it. However, some employees may be preoccupied about how this technology can affect their jobs. Introducing the solution as a way to supplement their daily tasks is important.
Be transparent with your employees on how the technology is implemented, why you’ve made what choice, how it resolves issues in a workflow and the employees’ benefits in using it.
Are you ready for AI?
We hope that this blog has helped you understand what you need before launching an AI project. If you would like to discuss further about your journey towards applied AI let us know: sales@skim.it, our data experts will get in touch with you as soon as possible.
Originally published at https://www.skimtechnologies.com on June 5, 2019.
