Are You Preparing to Fail on AI?
Are You Preparing to Fail on AI? Here’s A Handy Checklist for Companies
By Michael Tae, Head of Strategy, Broadridge Financial Solutions
Artificial Intelligence (AI) is now everywhere in our daily lives. It makes new products and business models possible through deep analysis of mountains of data, predicting things that humans cannot foresee, and relieving employees of grunt work. For example, AI underpins the intelligent automation software in the logistics industry that is improving capacity planning. Pharmaceutical companies are working to improve the efficiency of research and clinical trials through AI systemic analysis of facts and extrapolations of likely outcomes. In the financial industry, AI technology locates better borrowers for loan companies, identifies money laundering patterns, and improves compliance reporting.
AI is still in its early days, but the technology is poised to generate serious value, contributing up to $15.7 trillion to the global economy by 2030 according to PwC estimates. Our recent AI Outlook Survey of the financial services industry indicated that 90% of companies are in some stage of AI adoption, or at least exploration, and 84% have already progressed beyond proof of concept with at least one development.
Despite its undeniable potential, often a crucial ingredient is missing within its application, whether it’s clean data, in-house talent, or the right culture. Sometimes the problem is more fundamental: Many firms are simply unsure what problem they are trying to solve. The technology is usually expensive, which means that failure can be costly—especially in the financial services industry, which is grappling with thinner margins.
Here are six questions that companies can ask themselves to avoid suffering that fate.
- Goal: Do I even need AI?
It is a mistake to throw cash at AI simply because everybody else is, especially if you struggle to see its benefit above much simpler solutions; experiencing the IKEA effect is not uncommon – where you mistakenly think that your product is valuable, but only because you have painstakingly built it.
A much healthier approach is to carefully outline a goal and then explore if AI would expedite the results. You should then ensure your AI budget matches your strategic goals, then define the metrics and milestones that you will use to measure its success. Fortunately, AI and Machine Learning are both tools that provide the ability to execute controlled pilots, where visibility on costs can be maintained and operational disruption can be minimized.
- Data: Does my AI have strong foundations?
It is a mantra among AI experts that any application of AI is only as good as the quality of its data. You need to collect clean data in the right formats, systems and quantity before your AI can do any deep learning, which may be a time-consuming and expensive process.
A good first step when assessing the role that AI can play in your organization is to draw up benchmarks on data quality, since only clean, abundant data will allow you to reap the benefits of the technology. Computing power is no longer a limitation to incorporating the technology. As companies migrate to cloud services, leveraging the data, analytics and Machine Learning stack of cloud providers would be the logical strategy.
- Partners: Can I do this in-house?
Financial players contemplating AI face a perennial dilemma: Should they build technology themselves – thus fostering the entrepreneurial spirit of a fintech — or ‘go outside’, by partnering with or buying fintech firms? The ultimate decision will vary according to company culture, but patient shareholders who will indulge the ‘fail fast and fail often’ mentality of a private, entrepreneurial fintech start-up are unusual in our industry. If you decide to go at it alone, it is critical to win buy-in from the company’s board to ensure that the firm hires talented data scientists and is committed to AI as a long-term investment.
- Talent: Who should I hire?
Even if you go down the partner route on AI, you will still need in-house expertise to stay on track. There will be huge competition for AI talent in the financial services industry within the next few years. The World Economic Forum estimates that industries will create 133 million jobs in AI and ML by 2022. Your organization will need data scientists, ML engineers, software developers, robotics experts and business intelligence researchers. The new recruits will ideally understand both data science and the financial services business itself, but you can also hire business experts to bridge the skills gap, or alternatively, teach operational experts to develop technical skills. Investing heavily in re-skilling your work force can break down barriers to change and ensure a growth path for talent within the firm.
- Buy-in: Are we all on the same page?
The ultimate goal should be an organization-wide understanding of AI, and the link between the technology and business opportunities. Executive buy-in to AI is just the first step – they can set the tone by engaging frequently on the topic. They will need to ensure that departments understand which areas of the company have greatest potential from technological change and articulate a sophisticated long-term strategy for AI.
- Patience: Am I ready for a long journey?
Companies should not doubt the utility of AI and that it has a role to play, but it is worth managing expectations around the technology. If you embark on this journey with the correct mindset– and that means understanding your goal and gathering clean data – then there is good chance that you can reap large gains. As stated above, establishing clear milestones that track progress and can keep stakeholders informed is a critical step to success, even if that success means failing quickly.
There is no doubt of the potential for Artificial Intelligence and Machine Learning. With the right preparation, you will not need to prepare to fail.
The new AI-based technology brings firms hard- and soft-dollar returns.
Companies need to view AI as a digital labor force rather than a solution, FIS CTO writes.
Exchange operator aims to catch more unusual price movements, trading errors and potential manipulation.
A small step for trade breaks may lead to giant steps for other applications.
Creators of autonomous technology won't escape regulators' reach.