‘…organisations enabling AI at the enterprise level are increasing operational efficiency, making faster, more informed decisions and innovating new products and services.’
So says a recent EY report, in language that’s characteristic of the fervour surrounding the use of artificial intelligence at work.
The claim that AI will have a direct impact on everything from finance to law and medicine is no longer outside the realm of possibility. Early adopters, often in the form of large corporations, have put use cases to the test. They’ve seen enough measurable success to back the kinds of bullish statements made by analysts at EY, and the above quote is representative of a trend towards eager anticipation of AI’s potential.
What steps need to be taken to achieve enablement, though? It’s exciting to think about the potential of the technology, but there’s currently a knowledge gap between ideation and implementation. Larger corporations have the bandwidth to experiment and learn, but those working in law, healthcare and other regulated, safety critical sectors – fields with far less margin for error – often don’t have that luxury. It’s a barrier to adoption that can only be avoided using a formalised implementation process.
Preparation is a necessity if AI is to deliver on its promise and foundations must be laid. We’ll cover what needs to be done, and why – but first, what exactly is it that’s being enabled?
What systemisation makes possible
AI can – and likely will – touch aspects of almost every industry. It’s not limited to the well-known use cases that we’ve seen with Spotify’s recommendation system, or experienced with Siri, Alexa and Cortana. There’s a world of applications outside of the tech sphere, resulting in everything from the cost-efficiency and productivity boost you’d like to see no matter your industry, to the more focused impact that it can have on fields like medicine and law. We’ve gathered some examples.
The universal benefits
There are certain tasks that AI can perform that will appeal to the C-Suite and to managers no matter their field.
HR is one of the areas that stands to experience significant change. Software like Entelo, for example, is used by marketing giant HubSpot to handle the thousands of applications they receive, and to seek out viable hires. The process is a joint effort between man and machine, allowing professional recruiters to review results and ensure that the selection process is fair and is guided by the needs of the organisation. That’s an important distinction considering the recent controversy surrounding the use of AI in recruitment.
It’s already made a huge impact. Becky McCullough, HubSpot’s director of recruitment, credits the tech with setting ‘new benchmarks for response rate’, and says that ‘it has put more rigour [into our process].’
AI is also impacting the customer-facing side of the equation. Increasingly sophisticated chatbots like Watson Assistant use language recognition and machine learning (ML) to deliver near-human interactions, allowing customer service staff to focus more of their attention on complex problem-solving that the technology can’t yet handle.
ML software is able to process structured information on an exponentially larger scale than a human brain. This enables it to scan and highlight relevant legal documents at a rate and level of accuracy that even the most detail-oriented, efficient law professional could only dream of. It’s already being put to use, with specialist programmes like ROSS Intelligence tailored to perform legal research.
Once the data is processed, it’s a matter of collaboration between humans and technology, as interpretation comes into play. Lawyers are freed from busy-work and are able to spend more time doing valuable work for their clients. With the average lawyer spending just 29 percent of their time on billable work, the introduction of machine learning could completely revolutionise the current state of play by maximising profitability.
‘AI will allow doctors to become more human.’ – Dr. Simon Eccles, Chief Clinical Information Officer for Health and Care at NHS England
In much the same way that AI and ML have the potential to free law professionals from administrative work and repetitive research, doctors could soon be able to spend more time focusing on the creative, human side of their careers.
With the rise of explainable AI (XAI) comes increased trust in the technology, and in its ability to make life-altering decisions. While ML-led AI still has a ways to go before it’s free of the ‘black box’, there’s demand for this to change. Healthcare is understandably a trust-sensitive industry, and it’s also one that could see the most profound benefits from the introduction of artificial intelligence. AI’s ability to process thousands of patient records and its predictive capability make it one of the most valuable tools available to doctors that are already stretched for time. With NHS GPs spending an average of just over nine minutes with each patient, reliable time-saving software can help better the quality of primary care at the point of delivery.
A report published by the Academy of Medical Royal Colleges suggests that AI could ‘[raise] minimum standards and [reduce] unwarranted variation’ in assessment and treatment, and could ‘improve access to healthcare, providing advice…in real-time to patients…’.
The major hurdle the Academy identifies?
Explainability and trust.
Collaboration unlocks potential
Writing in Harvard Business Review, Accenture’s H. James Wilson and Paul R. Daugherty of make this point about AI’s capabilities in the workplace:
‘[AI] can amplify our cognitive strengths; interact with customers and employees to free us for higher-level tasks; and embody human skills to extend our physical capabilities.’
It’s further high praise for the potential of the technology, but the benefits are all qualified by a ‘can’. The variable is the level of human input. Wilson and Daugherty’s research makes it clear that without human-machine collaboration, AI is nowhere near as effective. Systemising AI use means presenting the software with the path of least resistance, so that it can achieve everything it promises to.
The benefits of systemising AI at work
Systemising the use of AI before it’s implemented means being proactive, not reactive. A formalised process eliminates confusion further down the line, removing organisational hurdles that could prevent your business from reaping the rewards that AI offers.
What does systemisation actually represent in terms of the benefits, though? In short, it means:
- Streamlined processes and implementation.
- Increased profit margins thanks to the time and money saved by clearly-established guidelines for AI use.
- Cross-organisation alignment on AI use (i.e. when collecting and entering data).
- The ability to scale the technology responsibly and within an established framework.
- An easier process for measuring the success of AI implementation.
- A clear set of expectations for the technology, defined at the outset.
- A clear ‘chain-of-command’ within the organisation when it comes to managing the technology.
Think of systemisation as the facilitator that makes it possible for AI to do its job properly. Without it, there’s a significant risk of missing out on the full ROI of the technology.
Systemisation in practice
Understanding AI’s potential
Before taking any other steps, it’s important to acquaint yourself with AI and ML as a whole. Getting a big picture of the technology’s current capabilities provides the context with which you’ll be making decisions further down the line, and the areas for adoption that you’ll be able to identify.
There are plenty of free resources online to take advantage of. The Ditto.ai blog, for one, is a great place to start when it comes to learning about some of the bigger questions being asked in the industry, as well as the practical applications of the technology.
Learn with Google AI is a free, one-stop shop for a more formalised approach to AI education, with broader courses like ‘AI for Social Good’ offered alongside more specific tracks like ‘Introduction to Machine Learning Problem Framing’.
Finding areas for improvement
The ‘scattergun’ approach doesn’t work when it comes to sustainable adoption of artificial intelligence. It’s important to identify specific problems that you’d like the technology to solve, so that the implementation is contained, and its ROI is easily measured. A focused approach will prevent confusion by encouraging an efficient rollout. Setting parameters from the outset and being as specific as possible about what you’d like to achieve will keep your AI team on-track and accountable. It’s incredibly valuable to record everything, and to put numbers to your goals. Introducing AI is about making your business more efficient, and in order to measure its success, you’ll want well-informed goals to aim for.
Identify the challenges
It’s helpful to take inspiration from existing success stories, but your organisation will have its own set of challenges to contend with. Invest time in defining the hurdles to adoption unique to your industry and company, and consciously work these into your implementation plan.
It might be the case that you have the skills required in-house, but it’s often the case that there’s a gap in capability that will need to be filled before adoption can happen. This can mean hiring AI experts, or it can mean upskilling your existing team. Whatever the case, it’s an essential early step that must be taken. Create a designated group to manage the project, with representatives from both IT and the business-at-large, if it’s possible. As Luke Tang of the Global AI+ Accelerator Program says:
‘It’s…important that expertise from both sides – the people who know about your business and [those] who know about AI – is merged on your pilot project team.’
Create a budget and timeline
Once your team is gathered, the next step is to craft a detailed budget and timeline for the project. Understanding the level of investment you’re able to make – and the ROI you’d like to see – will make it easier to narrow your focus and determine what technology you’ll be implementing. Involve the entire team in the process, making the most of the business and AI expertise to create a holistically-defined investment plan.
A timeline is equally important, as AI isn’t a ‘set-and-forget’ technology. There should be clearly-flagged dates for analysis and engagement with the programme, and points at which the project’s success should be measured. Without a predetermined set of milestones, it can be difficult to keep the software accountable, and to make improvements where necessary.
Prepare your existing assets
The phrase ‘garbage in means garbage out’ frequently surfaces in the world of AI, and for good reason. The technology’s ability to offer insights and increase efficiency is only as good as the information it’s given. If data isn’t standardised and organised ahead of time, the AI’s entire knowledge base is marred. Further Ditto AI’s patented methodology helps systemise tacit and unstructured information, enabling organisations to take advantage of the knowledge residing in their employees and processes.
Investing time in optimising your organisation’s existing data – and regulating future records – before implementation will save countless hours and exponentially increase the software’s chances of delivering real, valuable success.
ML programmes often work best after being ‘trained’ with existing situations and the information associated with them. A basic example comes with Spotify’s recommendation system – the more data it gathers on the kinds of songs you like, the more accurately it can suggest others you’ll probably enjoy.
The same goes for more complex ML systems used in the finance industry – JP Morgan Chase’s technology has learnt from thousands of transactions and investments in order to make historically-informed decisions.
In much the same way that it’s important to prepare your existing assets, it’s crucial to ensure that your ‘training’ information is optimised for use with the software you’re planning to introduce. This is where the experts come in.
Wilson and Daugherty’s aforementioned research identified five key areas that organisations with effective AI programmes actively pursued. The more active businesses were in said areas, the more rewards they saw:
- Reimagining business processes
- Embracing experimentation and employee involvement
- Actively directly AI strategy
- Responsibly collecting data
- Redesigning work to incorporate AI and to develop relevant employee skills
What do all of these activities share? They’re all focused on systemising the use of AI at work. Take an active role in collaborating with artificial intelligence and formalising its use, and you’ll be far more likely to see some of the evidence that backs the hype.