Why professionals should not depend on AI – Updates on AI & Technology by Outreinfo

When asked to name a company using AI, people often list technology powerhouses. However, leaders at legacy organizations in other industries may feel it’s beyond their capabilities to transform using AI. Since AI is relatively new, all successful companies had to accomplish the same tasks: putting people in charge of creating AI, acquiring data, talent, and investments, and aggressively building capabilities.

At many organizations, AI initiatives are too small and tentative to add economic value. A 2019 survey found that 7 out of 10 companies reported minimal or no impact from their AI efforts. However, research has identified 30 companies and government agencies that have successfully adopted AI by going all in and reaping the benefits. These organizations took 10 actions to become successful AI adopters.

To get value from AI, organizations must rethink how humans and machines interact. Focus on applications that change employee performance and customer interaction. Consider deploying AI across key functions to support new processes and data-driven decision-making. AI should drive new offerings and business models, eventually transforming every aspect of the business

To fully achieve AI transformation, businesses must avoid piecemeal efforts and tackle all 10 tasks identified in this article. Examples detail how some organizations succeeded. Your business may handle the tasks differently or approach them in a different order.

Ambitious companies have a specific sense of how they want to apply AI. Identifying and developing transformational AI requires a clear objective, such as improving process speed, reducing costs, or becoming better marketers. We recommend identifying one well-defined, overarching objective to guide AI adoption.

When Deloitte’s audit and assurance practice developed Omnia, a proprietary AI platform, the guiding principle was to improve service quality globally. Creating a global tool required addressing differences in data regulation, privacy, audit processes, and risk management. Extracting relevant data and loading it onto an auditing platform can be labor-intensive due to differing data structures. Developing a single data model that would work across clients and regions presented unique challenges.

Envisioning Omnia as a global tool before it was created allowed Deloitte’s developers to focus on standardizing information from different companies in different countries. This was a huge undertaking that would have been even more challenging later in the development process.

Collaborative Partnerships for Enhanced Results
Building Omnia required Deloitte to monitor technology start-ups to find solutions that fit their needs. Strong partnerships are necessary for AI success. Deloitte worked with Kira Systems to extract contract terms from legal documents using natural-language-processing technology. Another partner, Signal AI, built a platform to analyze financial data and identify risk factors. A recent addition to Omnia is Trustworthy AI, developed with Chatterbox Labs, which evaluates AI models for bias.

analytics initiatives
Most successful AI adopters had significant analytics initiatives before moving into AI. Mastering analytics is crucial to AI adoption and requires a commitment to using data and analytics for most decisions. This means changing customer interactions, embedding AI in products and services, and automating tasks and processes. To transform their businesses with AI, companies must have unique or proprietary data to differentiate their machine-learning models and outcomes.

Revolutionize Human-Machine Interaction for AI’s Full Potential
Seagate Technology has been using sensor data to improve manufacturing processes, including automating the visual inspection of silicon wafers. An automated system allows machines to find and classify wafer defects, saving millions in inspection labor costs and scrap prevention. Visual inspection accuracy now exceeds 90%. Data is the foundation of machine-learning success, and acquiring, cleaning, and integrating the right data is the biggest obstacle for scaling up AI systems. Pursuing new sources of data for new AI initiatives is also important.

Build an Agile and Adaptable IT Infrastructure
Deploying data, analytics, and automation across enterprise applications requires a technology infrastructure that can communicate with other IT environments. Integrating software from outside a traditional data center can be time-consuming and expensive. A flexible IT architecture makes it easier to automate complex processes. If you can’t develop such an architecture, you may have to partner with a company like Microsoft Azure, AWS, or Google Cloud. Capital One modernized its culture, operating processes, and technology infrastructure by moving to an agile model, building an engineering organization, hiring for digital roles, and moving its data to the cloud.

Capital One built its cloud architecture in partnership with AWS. Before the move, executives had to reimagine the future of banking. Digital channels produced more data than in-person interactions, giving the bank an opportunity to better understand customer interactions. Shifting to the cloud made strategic sense because it would drive down data storage costs, which have dropped from $2 million per gigabyte in 1960 to as low as 2 cents by 2017 thanks to cloud storage.

Incorporate AI into Current Work Processes
Inflexible business processes can be as limiting as inflexible IT architectures. Successful companies integrate AI into the daily workflows of employees and customers. Determine which workflows can benefit from AI speed and intelligence and begin integrating AI into them. Avoid trying to cram it into workflows that wouldn’t benefit from machine speed and scale. Workflow integration requires a specific plan of attack and on-the-ground knowledge of processes. Line employees have an ideal perspective for determining which processes can benefit from AI and how they can be improved.

Some branches of the U.S. government identified specific tasks and workflows ideal for AI speed and scale. NASA launched pilot projects in accounts payable and receivable, IT spending, and HR. The Social Security Administration used AI to address heavy caseloads and ensure accuracy in decision-making. The Department of Veterans Affairs implemented AI chatbots during the Covid-19 pandemic. The Transportation Security Lab is exploring ways to incorporate AI in TSA screening. The IRS is using AI to test which notices are most likely to induce payment.

Develop Holistic Solutions Across the Company
Once you’ve internally tested and mastered AI across a specific workflow, aim to deploy it aggressively throughout the organization using a unified approach that can be replicated. Cleveland Clinic has “AI popping up all over the place,” facilitated by worker-led efforts to develop and deploy AI with executive-led governance. The effort has been driven by a cross-organizational community of practice anchored in enterprise analytics, IT, and ethics departments.

Like most organizations beginning aggressive AI transformations, Cleveland Clinic faces challenges with data and analytics. Hospitals have less data, which is less likely to be clean and well-structured. Cleveland’s data has quality issues and is captured and recorded in different ways. Donovan’s group makes data preparation part of every AI project and provides useful data sets to all AI projects. The clinic uses AI to assess risk in population health, building predictive models to prioritize care and proactively schedule preventive care. They also work to identify patients with problematic living or working conditions that affect their health.

Establish Effective AI Governance and Leadership
Putting someone in charge of AI deployment makes transformation easier. The best leaders are aware of AI’s capabilities and implications for their companies. The greatest challenge is creating a culture that emphasizes data-driven decisions and makes employees enthusiastic about AI’s potential. Without this culture, AI advocates won’t get the resources or support they need. Leaders should be familiar with information technology, work on multiple fronts, and invest in exploring, developing, and deploying AI. Commitment to AI must go deep into the organization, with upper, middle, and frontline managers all on board.

Establish Centres of Excellence for AI Development
AI and analytics heads often promote the technology’s value to other managers. Decision-makers should support AI projects with funding and time, and use AI in their work. Educating them on AI’s function, appropriateness, and commitment is important. While not all employees need AI training, some do, and more is better. Successful companies need talent and training in AI, data engineering, and data science.

When Piyush Gupta became CEO of DBS Bank in 2009, it had poor customer service ratings. Gupta invested $300 million yearly in AI experimentation and allowed business units to hire data scientists. The HR head created a group to pilot AI applications, including JIM, which predicts personnel attrition and helps recruit qualified employees. DBS now has twice as many engineers as bankers, working on emerging technologies and AI projects. The bank’s culture improved, earning it recognition as the world’s best bank and high credit ratings. Gupta was named the 89th best-performing CEO in the world in 2019.

Ensure Continuous Investment in AI
Aggressively adopting AI is a major decision that can cost large enterprises billions of dollars. Successful AI adopters see it as the cost of committing to ambitious enterprise-level adoption. After seeing early benefits, companies find it easier to spend on AI-oriented data, technologies, and people. CCC Intelligent Solutions, for example, spends over $100 million yearly on AI and data. Founded in 1980 as Certified Collateral Corporation, it provided car valuation information to insurers. CCC evolved to collect and manage data, establish relationships in the automobile insurance industry, and make decisions with data, analytics, and AI. Led by Githesh Ramamurthy for 23 years, CCC has grown to approach $700 million in annual revenues.

CCC’s machine-learning models use historical claims, images, and data on automobile parts, repair shops, collision injuries, and regulations. It also has telematics and sensor data from over 50 billion miles. CCC provides data and decisions to an ecosystem of insurers, repair facilities, parts suppliers, and automobile manufacturers. Its goal is to link these organizations to process claims quickly. Transactions take place in the cloud, where CCC’s systems have been based since 2003. They connect 30,000 companies and 500,000 users and process $100 billion in transactions annually. Reaching this point has been expensive and time-consuming.

Explore Diverse Data Sources Continuously
Large companies typically have no problem gathering data, but AI strategies depend on the data available. More accurate, structured data that can be immediately applied to AI models is ideal. Integrating client data was challenging for Deloitte. Capital One had strong data but needed a flexible IT architecture to store and use it. CCC accumulated data from its first business model and was prepared for a shift to AI. Its transition to an AI-oriented business was solidified when it learned to use a large amount of new data.