Four Innovative Approaches & a heads-up for Leveraging Generative AI in Research & Consulting

Consultants, Researchers and Academics Can Use Generative AI to Supercharge Their Productivity

‍

Generative AI refers to algorithms that can be used to create new content or synthesize existing content from large quantities of audio, code, images, text or data sequences.

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

Regardless of the field or industry you work in, project lifecycle has four phases: scoping, planning, executing and delivering—Gen AI can help throughout. Large language models (LLMs) are here to assist—not to replace you. A change in mindset is essential.

‍

‍

‍

‍

It’s nearly impossible to scroll through daily headlines without encountering commentary on generative AI (Gen AI)—the latest frontier of artificial intelligence.

Undertaking relevant thorough and rigorous research can be a real struggle in both academic and business world. Academics, researchers and consultants must employ innovative research methods, carefully analyse complex data and then skilfully and clearly write, all while keeping the interest of a broad audience in mind.

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

‍

“Not long time ago, researchers, consultants or academics ability to build a research framework, collect, analyse and present insights to tackle a certain issue represented a big and tedious challenge—Gen AI can assist.”

‍

It’s nearly impossible to scroll through daily headlines without encountering commentary on generative AI (Gen AI)—the latest frontier of artificial intelligence.

Undertaking relevant thorough and rigorous research can be a real struggle in both academic and business world. Academics, researchers and consultants must employ innovative research methods, carefully analyse complex data and then skilfully and clearly write, all while keeping the interest of a broad audience in mind.

“Not long time ago, researchers, consultants or academics ability to build a research framework, collect, analyse and present insights to tackle a certain issue represented a big and tedious challenge—Gen AI can assist.”

As someone uses Gen AI in my daily life over the last three years, years before OpenAI GPT, in professional work and beyond, I have spent a lot of time thinking about the potential benefits and challenges—I must say that AI saved me a few times so far and therefore I can discuss the impact of AI on research careers. Let’s not forget that AI is a machine, while it excels in certain tasks as it is trained to possess certain skills, it cannot replicate the passion and unique personality that drivers research motivation; however, what it can do is help spark our genius.

Below, I offer several ways AI can inspire our research, from brainstorming, analysing data, verifying and presenting findings.

1. Leverage Gen AI in Brainstorming

Utilising Gen AI, specifically advanced models like ChatGPT-4 and more recently Gemini Advanced , has revolutionised my daily work whether researech or consulting. These tools offer great assistance in writing, creating graphics, analysing data and internet browsing, mimicking human-like capabilities.

I engage with Gen AI in a conversational and natural way, tailoring my prompts to the specific context of my work. For example, in my recent research paper, I’ve utilised prompts like:

• "I am thinking about [topic], but this is not a very innovative/pratical idea. Help me find industry reports and innovative papers and research ideas from the last 5 years that has discussed [topic]?"
• "What current topics are being discussed in the business press? or when I am working on an engineering projects, I would say what are the latest trends in lightweighting technologies? “
• “Make a table of methods that have and have not been used related to [topic] in recent research or consulting practices?”

The objective is not just to generate a single sufficient prompt but to refine the AI's output into robust and reliable results, validating each step in the process as a thorough scholar would. Sometimes because AI has been trained on multi-disciplinary data, it might spark an idea for further exploration, or it might not be helpful at all. However, asking these questions often aids in overcoming obstacles associated with difficult research problems.

After the prompts, there's still a significant amount of work to be done. Yet, having a Gen AI companion expedites the process. Prompts like, “Explore uncovered areas in engineering applied research commercialisation” and “Explore uncharted areas in university spinout strategy and business models” have led to the identification of promising avenues for future research projects. Should I need more novel ideas, I simply prompt for suggestions or ask to expand on a particularly interesting point, which often yields interesting project proposals.

For consulting projects, I leverage prompts such as:

• “You are a senior consultant, provide a list of the latest practical frameworks on theories on market entry in the EdTech sector in the US.”
• “Create a table of the latest case studies on business strategy in international markets.”

This method of interacting with AI not only as a tool but as a brainstorming partner has significantly enhanced my research and consulting efforts, paving the way for more innovative and efficient outcomes.

2. Leverage Gen AI in Data Collection & Analysis

While Gen AI has many flaws, it provides iterative feedback significantly enhances its effectiveness, turning it into a powerful tool that brings clarity and insight to ideas. My approach to using AI doesn't rely on a static set of prompts for data collection. Instead, I've found its strength in aiding with coding tasks, including the writing and debugging Python scripts across various programming languages, to be invaluable.

For example, ChatGPT-4 has been instrumental in building applications for web scraping and data collection, offering codes that are easy to comprehend, identify errors and resolve them. This marks a stark contrast to the pre-AI, where troubleshooting software issues consumed more of my time than coding itself. I can pose questions like,

• What's the optimal strategy for data collection on [topic]?
• Which software is best suited for this task?
• How can you assist in acquiring this data?
• What coding approach should I employ for data collection?
• What is the most effective analysis method?
• As a critical reviewer, what additional aspects would you consider?

Though initial attempts may not always return helpful outputs, starting the process without AI is more challenging.

‍

‍

In situations where AI's responses do not meet expectations, I engage in a feedback loop, posing questions such as, "This attempt was unsuccessful. Here's my code; could you identify the problem?What led to the failure of this method?" or "This answer is not accurate. Could you propose two alternative strategies for achieving the desired result?" Occasionally, AI might suggest data that is unattainable or not useful. In such cases, my response is, "This data is challenging to access; do you have suggestions for accessible alternatives?" or "This data doesn't seem accurate; can you recommend innovative data sources or places to find the correct data?"

‍

“Integrating AI into the data gathering and analysis process not only opens up avenues for valuable insights and solutions to complex problems but also encourages the development of more efficient coding practices and the exploration of alternative data sources. This integration significantly enhances the efficiency of the research process.”

‍

3. Leverage Gen AI in verifying results & promot transparency

AI's capability to document the research evolution offers a digital audit trail, enhancing the credibility and reproducibility of findings. This transparency is crucial in a field where methodical rigor is important. A key advantage of this digital documentation is its role in improving transparency by providing traceable evidence of the research methodology, thereby bolstering the credibility of findings through a demonstration of the systematic approach undertaken.

For instance, while developing code to retrieve data from an external source, I posed questions to ChatGpt such as, “Can you identify any issues with this software program?” and “What output will the program generate?” If the code was inefficient and uses excessive memory, leading to prolonged execution times. Probing AI for a more streamlined and efficient coding solution resulted in an alternative that notably enhanced data processing speed.

What excites me is the potential to simplify the process to replicate my work. Documenting each iteration is a time-consuming task often bypassed by many in the field. However, with the assistance of generative AI, simplifying these documentations for better comprehension becomes feasible. Queries to AI might include:

• Provide step by step guide to replicate the work for enhanced clarity.
• Can I reproduce the findings using other methods.
• Provide a detailed depication of the study’s workflow.

For qualitative data analysis, questions could be, “Can you pinpoint discussions of this idea within the text and tabulate them for easier comprehension?” or “Identify texts that challenge these findings and discuss the conditions leading to these divergent examples.”

I usually ask AI to compile a database of all prompts leading to the derived results and data. This approach not only facilitates validation but also significantly elevates the research's reliability and integrity.

4. Leverage AI in Ancipating & Responding to Feedback

To predict potential feedback, I ask Gen AI questions like, “From the perspective of a critical client prone to highlighting issues, what weaknesses can you identify in my work? and how might I address these weaknesses?” This strategy allows me to scrutinise my work for any logical or analytical gaps and make necessary adjustments before presenting my findings to audience. Identifying potential issues early on, especially in the competitive and fast-paced environment of scientific research and consulting, proves to be both effective and efficient.

Following the actual feedback, I turn to Gen AI again to gain a clearer understanding of their expectations. By asking, “Can you help me pinpoint the main points of this review, arranging them from the simplest and quickest to address to the most complex and time-intensive?” I find the review process not only becomes more manageable but also more enjoyable. Gaining a comprehensive view of what the client seeks significantly simplifies the response process.

Be Mindful of Gen AI’s Capabilities, Limitations & Potential Risks

As professionals, it's crucial to adapt to and innovate alongside technological advancements. LLMs hold the promise of propelling research forward, catalysing breakthroughs that extend the boundaries of what's currently achievable. However, engaging with Gen AI entails confronting its inherent risks, which are substantial and warrant thorough consideration:

• Accuracy: Gen AI has the propensity to generate plausible but fabricated responses “hallucinations” or exhibit inadequate abstract reasoning capabilities. What might appear as high-quality answers, it could lack practical value. For example, a consultant advicing on engine component replacement faces a much higher standard for accuracy than someone seeking general troubleshooting advice.
• Security: Vulnerabilities to covert manipulations, such as backdoor attacks, pose a significant threat. There have been cases where malicious entities have haijcked AI models for spreading misinformation, data theft or fraud.
• Privacy: There's a risk of inadvertently exposing sensitive or confidential information through public LLM interfaces. It's imperative to ensure data fed into LLMs is carefully segmented, mirroring existing access restrictions, especially when handling private or top-secret information.
• Fairness: Similar to conventional AI systems, Gen AI can produce biased outcomes or be exploited to circumvent deliberately implemented safety measures.
• Legal Concerns: The possibility of infringing on intellectual property rights, committing copyright breaches and encountering liabilities due to misuse presents a legal quagmire. The legal framework governing the use of general AI outputs remains murky, with regulatory bodies across various regions still deliberating on appropriate oversight mechanisms.

Understanding and navigating these aspects of Gen AI is essential for harnessing its benefits while mitigating potential risks.

‍