Five key steps to successfully implement AI in your business
Machine Learning (ML) and Artificial Intelligence (AI) are undoubtedly the hot topics of the moment in technology, and the driving force of most of the big technological breakthroughs of recent years. Legendary tech giants such as Google, Facebook, and Amazon, along with thousands of large and small startups across the world heavily rely on AI to design, develop and market their products. Yet, relying on AI technologies and implementing them to develop innovative business strategies and products is no longer a prerogative of tech companies only. Indeed, thousands of companies operating in a broad range of industries — including retail, agriculture, medicine, and real estate — have adopted and implemented AI technologies to improve their business, keep track of the
and enhance the appeal of their brands. Switching from more traditional ways of conducting business to implementing AI projects can surely be beneficial to many industries and companies; however, it is no walk in the park. Implementing AI projects without developing a thorough and strong AI strategy and without fully understanding what AI is and what it can be used for may lead to huge loss of your money and time. This article will discuss a few major steps that any company that would like to implement one or multiple AI projects to their existing systems should consider taking before deciding to make a decision that will dramatically revolutionize their way of doing business forever.
Step 1: Make sure you know what AI can do for your business and how to leverage that.
“The first thing you have to do is figure out what AI can actually do for you as a company,” said AI and ML consultant Adam Geitgey in an article published by the online tech magazine TechRepublic in April 2019.[1] Indeed, it’s a very common mistake to be wanting to use AI simply because it’s popular, and jumping into software development before understanding the problem you need to solve. Therefore, the company’s leadership should try to answer six broad questions before getting their hands dirty with their first AI project:[2]
1. Why do we want to use AI for our existing system?
2. Do we have the resources, in terms of money, time, computational power, and manpower, to develop an AI project?
3. Will AI give our company a concrete advantage over our competitors if we apply it to the products we sell?
4. Do we want to use AI to build better products, or to get products to go to market faster?
5. Do we want to use AI to become more efficient or profitable in ways beyond product development?
6. Do we want to use AI to mitigate some form of risk?
Moreover, the company’s leadership should know both what the superpowers and the
Achilles’ heel of AI are.
Currently, AI can do three things well: automate process, create new products, or improve existing products. Nonetheless, C-suit executive also should be mindful that AI still struggles with some more complex tasks, such as context understanding.
Thus, the leadership of a company should have a clear idea of why they want to use AI, what problem(s) it could potentially solve, how it can be implemented to make their product better and more marketable, and what AI’s major strengths and drawbacks are. The company should have in mind specific use cases in which AI could solve business problems or provide demonstrable value.
Getting an answer to these questions might not be an easy task for companies that have not dealt with AI before. That’s why it’s important for companies’ leadership to familiarize with AI before making strategic decisions involving it. Companies could take advantage of the plethora of websites and online resources available to help people familiarize with the basic concepts of AI. In addition, companies could also hire an AI consultant, or contract out an AI innovation services provider such as TechCode Accelerator. Finally, every company that wants to implement AI projects to their business should perform an IT assessment as most companies are weighed down with old legacy systems, which can make it difficult to implement an AI project. Therefore, there must be a realistic look at what needs to be built to have the right technology foundation, which can be costly and take a significant amount of time.
Source: CIO
Step 2: Start small and no rush!
A common rookie mistake for non-AI companies that try to implement AI to their business is wasting time and money in seemingly glorious AI projects that end up being too complicated, expansive, or simply unfeasible. It’s best to be slow and specific about embarking on AI or ML projects. Therefore, rather than going all in, your company should begin by picking a small project that could benefit from cognitive technology, and use a smaller, less transformative toolset to attack it.[3] “It is more important for your first few AI projects to succeed rather than be the most valuable AI projects” claims Andrew Ng — founder and CEO of the AI solutions provider Landing AI — and the first project should
be technically feasible.[4] Ng also states that the first project should have “a clearly defined and measurable objective that creates business value.”[5]
Another common mistake is that many companies expect AI projects to magically solve problems in a short time period. Yet, AI projects usually start bearing their fruits after 6–12 months and might be completed in 2–3 years. Executives should keep that in mind when they plan their AI strategy, but they also need to be sure they can deliver at least some results within the first year. It is indeed important for your first AI project to deliver some results within a year to gain momentum and to keep stakeholders engaged and supportive.
Source: youcanmarketonline.com
Step 3: Create an AI team — training your employees in-house or hiring new professionals with a strong background in AI?
If your company is considering implementing an AI or ML project to its existing system, you need to make sure your company has a dedicated AI team. If it doesn’t, your company must build one as having a team of people with strong background and expertise in data science, software engineering, and coding is paramount to properly tackle AI’s complex problems and challenges.
When it’s time to create an AI team, the dilemma is usually the following: should the company create a new team by hiring people with relevant background and work experience in AI or should it build an in-house team by training the existing employees? There are currently various school of thoughts on this matter and it seems that the data science community has not agreed on a unanimous answer yet. For example, Andrew Ng argues that building an in-house team is better, as AI feeds off domain knowledge, which can be hard to find in certain industries.[6] Outsourced employees likely do not know your data, infrastructure, and problems as well as the company’s employees. Ng claims that with the rise of digital contents, including open online courses and YouTube, “it is more cost effective than ever to train up large numbers of employees in new skills such as AI.”[7] Nonetheless, other experts in the field argue that in a technical-skills-heavy environment such as AI, it’s necessary to have people who dedicated their whole professional career to solve ML and AI problems, and that companies should
definitively onboard these people if they do not employ any.[8][9] Therefore, the answer to the above question is probably “it depends.” It depends on the many variables this intricate AI game is made of. It depends on the size of the company, its budget, its existing line of business, and on the product(s) they want to apply AI on. For example, if your company already has some employees with technical, IT, or similar expertise, it might be easier and convenient to train them in AI. Conversely, if your company doesn’t have any employee who is tech savvy or someone who has a minimum technical background, it’s probably better to hire new people with a stronger technical background. If the company has the necessary resources, probably the best solution would be training an in-house team and hiring new employees with specific technical background to complement each other.
There is, however, one point the AI community seems to agree on. Regardless of how you decide to build your AI team, all the existing employees at every level needs to be trained to work in an environment where AI becomes a major player. They don’t necessarily have to have the technical expertise required; however, they need to become used to work with tools that they have never used before. Senior executives are the ones that needs to be trained first, as they are required to prepare, motivate, and equip the workforce to make a change. Hiring or training a senior executive as a CIO, CTO, or CDO (Chief Data Officer) to lead the newly created AI team might also be useful to ensure appropriate project management.
Step 4: Develop your AI Strategy
Once you have a clear idea of the problem you are going to solve and of how AI is going to boost your existing business, you need to develop and implement an AI strategy to put your project into action. The core components for a solid AI strategy are the following:
Data: to develop an effective AI strategy you need to develop a good data strategy first. You need to make sure you can find enough data to feed your AI algorithm with. Additionally, make sure the data you find is relevant, otherwise your model won’t perform well.
Infrastructure: you need to build an infrastructure made of both computational power and manpower. You need to build an AI team as discussed above and you also need to be sure your machines have enough computing power and storage capabilities to effectively develop and run your ML models.[10]
Algorithm: another common AI dilemma is: do you develop your own
algorithms, or do you use existing ones? Having your own algorithm can lead to major benefits. For example, you can tailor your algorithm to the specific product you are marketing to make it unique and gain an advantage over your competitors. At the same time, there are clear risks and expenses. You need to have a highly-skilled AI team, develop and test models, and then deploy the system. That’s why many companies decide to use existing algorithms and apply them to their system. In this case, one option could be to use certain features from existing software platforms, such as Salesforce.com’s Einstein. Alternatively, you could build your algorithm from open source software.[11] There is no shortage of open source cognitive software. Companies such as Google and Microsoft have recently released open source machine learning or deep learning algorithm libraries. Since the software is free, this approach offers the lowest software costs, although you still need a highly-skilled team that can implement the algorithm.
Source: VentureBeat
Step 5: Understand AI ethics and safety
AI is a powerful tool that has the potential to make a substantial positive impact for companies, communities, and society. Yet, it’s extremely important to understand that AI could be harmful and dangerous and therefore your company should make considerations of AI ethics and safety a high priority. These harms rarely arise as a result of a deliberate choice; most AI developers do not want to build biased or discriminatory applications or applications which invade users’ privacy. The main ways AI systems can cause involuntary harm are: misuse of the software, questionable algorithm design, and unintended negative consequences of the model.[12] The main way to reduce the risk of involuntary harm is simple: you and
your colleagues need to be aware that AI can in fact cause harm if not handled properly. In addition, your company should adopt some measures that can prevent involuntary harm from happening, such as establishing ethical building blocks for your AI project and creating and implementing a framework of ethical values that all the employees must follow.
Lastly, as argued by Andrew Ng, if your company operates in a highly regulated industry — such as healthcare — you should develop a “compelling story” that explains the value and benefits your project can bring to an industry or society.[13] It is also critical that your company maintains direct communication and ongoing dialogue with regulators as you develop your project.
Source: Emerj
Conclusion
This article gave the reader some insightful tips and hopefully it provided a useful roadmap for people who would like to implement their first AI project to their existing system. AI is an extremely powerful tool that can really change your business for the better and scale it up to the next level. Nevertheless, AI is no silver bullet or panacea that can be applied to any company and magically turn it into a successful and profitable business. You
need to analyze thoroughly why and how you should implement AI to your existing business by carefully developing a solid AI strategy and by building a team of trained and skilled professionals that have the necessary expertise to tackle AI’s complex challenges. Moreover, you should be aware that working on an AI product requires daily dedication and efforts and yet it will take some time to bare fruits. Implementing AI is a marathon that has to be run slowly yet at a steady pace.
References:
[1] https://www.techrepublic.com/article/how-to-implement-ai-and-machine-learning/
[2] https://www.forbes.com/sites/tomtaulli/2019/04/13/implementing-ai-the-right-way/#219c91a557e4
[3] https://hbr.org/2016/10/7-ways-to-introduce-ai-into-your-organization
[5] Ibid.
[7] Ibid.
[8] https://www.techrepublic.com/article/how-to-implement-ai-and-machine-learning/
[9] https://www.pcmag.com/news/10-steps-to-adopting-artificial-intelligence-in-your-business
[11] https://hbr.org/2016/10/7-ways-to-introduce-ai-into-your-organization
[12] https://www.gov.uk/guidance/understanding-artificial-intelligence-ethics-and-safety