Navigating the complex world of AI

Opening a new world of opportunities.

Are we confused yet?

AI is a catch-all acronym for artificial intelligence (AI) that’s been around for more than two decades. In fact, mature tools, algorithms and technologies such as machine learning (ML), deep learning expert systems, robotics and natural language processing (NLP), can all be classified as AI technologies.

To clarify, machine learning is a subcategory of AI that consists of writing algorithms to recognize patterns and then training them to do so. This technique is broadly used in predictive analytics to support data-driven decision-making. Other forms of AI such as expert systems can operate without the need to train data models.

The current hype around AI is mostly focused on Generative AI (GenAI), which is based on fast-growing large language models (LLMs) and relies on machine learning (e.g., OpenAI and others). Today, emerging technology is at the root of how people and businesses communicate every day.

Opening a world of new opportunities

What if, instead of investing time in writing summaries after remote meetings, we could automate the generation and distribution of meeting notes to relevant recipients? What if following up on tasks and action items is seamlessly documented, while timely reminders are dispatched as needed?

What if the scope of meeting transcription transcends its conventional boundaries, so that in addition to transcribing discussions, we could seamlessly infuse internal and external contextual information related to the subjects at hand? What if marketing departments could automatically generate first drafts of presentations from text or meeting recordings, product specifications, documentation, brochures?

What if early versions of complex code were made available to developers to enhance, and then finalize?

What if combining email and calendar platforms, customer relationship management (CRM) systems and other enterprise data with outputs from ML sales predictive models could better equip representatives to close deals?

All these applications are not at all far-fetched; vendors are already announcing products with similar features.

Embracing opportunities while mitigating risks

In fact, businesses are in theory facing countless potential “use cases” or applications leveraging GenAI across many areas: service and operations, manufacturing, supply chain, sales and marketing, product/service innovation, finance, risk/fraud management and human resources.

Undoubtedly, adoption and penetration of such applications will be uneven; legal and regulatory constraints will have a significant influence on some industries’ ability to push ahead aggressively with AI innovation.

The value unlocked by the implementation of new GenAI use cases is incremental to the value already created by leveraging advanced analytics, traditional ML and deep learning.

GenAI may accelerate value creation in areas where:

  • Routine and repetitive tasks are performed.
  • The level of reliance on text and language is higher than average.
  • Contextual information is key to sales conversion.
  • Creative content generation results from significant collaboration processes.
  • Use of huge volumes of unstructured data can be enhanced.
  • Heavy human workload is attributed to risk and compliance requirements.

The language models powering GenAI aren’t trained using logic. Therefore, outputs from the technology cannot and should not be viewed as a trusted, reliable source of truth. In addition, opening the floodgates to employees could lead to serious concerns and liability exposure (legal, privacy, confidentiality, etc.).

Planning and executing an AI journey from strategy to successful deployment

At this juncture, business execs must address key strategic and tactical priorities:

  • Aligning AI strategy and business strategy
  • Ensuring the “data house” is in order, as it is the foundation
  • Assessing AI’s potential impact on sector/value chain (e.g., legal, regulatory constraints)
  • Ideating and prioritizing GenAI use cases
  • Devising a plan to pilot and scale-up SAFEly (secure, adaptable, factual and ethical)* and responsibly.
  • Assessing required capabilities to deploy successfully
  • Exploring and executing partnerships, if necessary
  • Building a change management plan (people dimension)
  • Detailing a risk management/mitigation strategy

We advise our clients to follow a step-by-step approach that allows them to enjoy the benefits the technology can bring while limiting risks to the organization:

Step 1  Baseline: vision, strategy, governance

Step 2  Prioritize and plan: pilot definition, change management planning

Step 3  Design and deploy: solution implementation, testing

Step 4  Industrialize: metrics, scale-up, ongoing management, change management implementation

Ready to get started?

At Mazars, our specialists can help you navigate the journey and safely deploy selective GenAI applications appropriate for your organization.

SAFE AI represents Mazars’ AI framework that focuses on creating robust system security measures and adhering to data privacy, compliance and protection standards. It also encourages the development of best practices for AI security and collaborative efforts with external partners and stakeholders.

The information provided here is for general guidance only, and does not constitute the provision of tax advice, accounting services, investment advice, legal advice, or professional consulting of any kind. The information provided herein should not be used as a substitute for consultation with professional tax, accounting, legal or other competent advisers.