What ChatGPT and generative AI can do for your business

One million users a week after its launch. 100 million in less than two months. ChatGPT, the generative AI chatbot developed by OpenAI, has certainly whipped up a storm - setting a new record for the fastest-growing online platform.

Perhaps you've already tried it out of sheer curiosity, or maybe you're waiting to see how it develops before jumping in. Wherever you stand, it's difficult to ignore this impressive new technology that’s making shockwaves across every industry and generating a lot of noise in the media.

While plenty of people have revealed online how they use ChatGPT for personal projects, how can businesses take advantage of it? Like most tech solutions, ChatGPT and other AI chatbots help entrepreneurs and employees save time in their daily work lives. The key to using AI effectively is learning how to ask it the right questions - and knowing its limits.

You’ll get fully up to speed by reading this article, but we’ll share a key takeaway now. Generative artificial intelligence should only be used as an assistant to speed up low-priority tasks. The technology can’t (yet) be relied upon to complete much of the creative, added-value work performed by humans.

So, how does generative AI work? What are the most interesting use cases and how do you write an effective prompt? Finally, what are its limitations? Let’s dive in.

At this stage, artificial intelligence can generally be described as analytical technology. It uses the internet’s wealth of information to categorize data and make predictions. ChatGPT, like similar platforms hitting the market recently, is known as generative AI - a new technology that’s able to synthesize online information into original content, be it text, images, videos, or audio.

Can AI really mimic how the human brain works? Not really. There’s a long way to go before the fantasy realm of “general artificial intelligence” (GAI) becomes a reality. For years, science-fiction writers have prophesized that this advanced form of AI would become capable of thinking and reasoning like a human being.

ChatGPT’s eerily convincing responses may have fooled many into believing the chatbot is sentient. However, its creators have assured us it isn’t capable of independent reasoning, and only functions according to human instructions.

In short, generative artificial intelligence does not actually know anything - it infers. Philosophical questions aside, the technology "learns" from billions of data sets to produce a coherent response to a user’s instruction, or "prompt".

Generative AI deduces the most relevant answer to a prompt, and is capable of expressing a human-like response based on the extensive data it’s been trained on. The potential of this technology seems limitless; it continues to learn with each new piece of content it absorbs, becoming more "intelligent" every day.

If ChatGPT was the first generative AI to emerge from the research labs and grab the media's attention, it's certainly not the only one out there. All the big names in tech have been working on this subject for years, mostly through open source. This explains why product releases have been picking up momentum in recent months.

Here are a few examples of the major players:

• Text generation: Alongside OpenAI’s ChatGPT, there’s also the chatbot HuggingChat, created by Hugging Face, or Google’s solution Bard (not yet available in France).
• Image generation: You’ve probably seen some spookily realistic images going viral on social media recently, showing the power of this technology. Midjourney, an independent research laboratory, seem to be leading this race, and OpenAI are again in the running with their product DALLE-2.
• Audio generation: Generative AIs can even create realistic sounds and audio. Google's MusicLM and OpenAI's JukeBox are useful options in this area.

Although generative AI has only been available to the general public for a short period of time, there’s already a variety of promising use cases for businesses. And as the technology progresses over the next few years, users will develop ever more creative and innovative applications.

Like other tech platforms on the market, generative artificial intelligence can help you optimize a wide range of operational processes: reduce the time taken to complete administrative tasks, give non-technical teams a helping hand, or get inspiration on almost any topic you could imagine.

The good news is that companies across all industries can take advantage of the technology, from finance and education to construction and retail. In addition, it can be put to good use by teams in tech, marketing, product, customer service, legal, etc.

It's impossible to cover every potential use case, but here are some examples to get the ball rolling:

• Automate administrative and project follow-up tasks
  Let's take the example of a construction contractor, who's in charge of managing a renovation project. With generative AI at their fingertips, they can automate a variety of tasks that previously had to be done manually, such as scheduling appointments, creating purchase orders, or sending invoices. The contractor can then focus on value-added tasks such as facilitating cooperation between the various stakeholders involved in the project. ChatGPT, or a similar chatbot, can also be used to optimize how resources are allocated, to track costs, and produce in-depth reports outlining where efficiencies can be made.
  
  
• Enhance customer and employee experiences
  Companies can easily use these advanced chatbots to handle basic customer queries, freeing up time for the customer support team to focus on more complex requests. Customers also get a faster response and can carry on with their day, satisfied that their issue has been resolved quickly. HR teams can put chatbots to work in a similar fashion, using them to answer routine questions from employees without delay.
  
  
• Analyse data and automate report writing
  One of generative AI’s key strengths lies in its ability to analyze, process, and make sense of data. It can detect underlying patterns and trends, and then summarize them in digestible language. This allows businesses to accelerate data analysis and make informed, data-driven decisions more rapidly than ever.
  
  

To sum up, generative AI provides useful assistance and helps speed up a wide variety of tasks, saving precious resources for entrepreneurs and their teams. Rather than replacing people - an understandable fear that’s been amplified in the media - AI can significantly enhance employees’ capabilities. In particular, AI makes highly sought-after technical skills, such as coding, more accessible for businesses.

Every day users are innovating with the technology and discovering new ways to apply it, paving the way for developers to create highly specialized software. GitHub and OpenAI’s Copilot, for example, pitch themselves as an "AI peer programmer" that can automatically complete developers’ code.

To get the best possible results out of platforms like ChatGPT, you need to provide them with the right information. This means writing effective prompts that specify exactly the kind of answer you’re looking for.

But what does an effective prompt look like? Ever since these generative AI platforms exploded onto the market, users have been experimenting with inputs to determine what works best. If you’re looking to craft the perfect prompt, a quick Google search will offer a huge variety of in-depth guides and tutorials on the subject.

Here's a quick introduction to writing an effective prompt. Keep in mind that the more precise your query is, the better the AI's response will be.

Your prompt should specify:

• The purpose of the query: what will you use the answer for?
• The expected format: are you expecting a complete article, an outline, or bullet points?
• The tone: do you want the text to sound formal, informative, or conversational?
• The context: what's the general background to your request?
• The target audience: who is the answer intended for?
  
  

You can refine your prompt further by adding:

• Limits: if you want the chatbot to produce an article, for example, you can set a maximum word limit.
• Keywords: specify any particular phrases or terms you want the chatbot to include - this is especially important for SEO content.
• Constraints: you can instruct the chatbot to exclude any sensitive topics from its response, if necessary.
  
  

Of course, you can specify your request with whatever information you deem relevant. Giving the AI as much context as possible is key - the more you guide it in the right direction, the better the result will be.

Artificial intelligence is undergoing constant research and development. What seems clear at this stage is that businesses should use AI platforms for assistance, not to complete tasks that require human creativity, reflection, and emotional intelligence. To put this in context, ChatGPT and Huggingf Face even include disclaimers on their homepages about the limitations of their products.

Therefore, it’s worth taking a moment to explore the technology’s limitations.

Let's start with the technical limitations that come with generative AI platforms like ChatGPT.

• The knowledge base is limited

  ChatGPT is built on a database that hasn't been updated since 2021, meaning that it's unable to answer questions about recent events. However, OpenAI claims that GPT-4 - the chatbots latest incarnation - will rectify this issue.
  
  

• AI can be wrong
  It appears that AI can also make mistakes. Many users have found that AI platforms are capable of making up certain concepts or details - a process described as "hallucination". It’s therefore crucial to check whether the response is indeed accurate, especially by combing through the source material.
  
  

• The response may be biased
  Generative AI platforms learn from billions of data sets. However, like plenty of online content out there, the source material may be biased. The danger is that AI can reproduce this material in its responses, without knowing the potentially offensive or inappropriate nature of the content.
  
  

• Produced content is based on existing material
  Since generative AI learns from the information fed to it, it's impossible to ensure the content it produces is 100% original. This means, in a sense, the AI is merely synthesizing and reproducing content that’s already out there.

Here's a roundup of the legal and security issues to keep in mind when using generative AI, which can have a big impact on businesses:

• All the data you share on the platform is used to train the AI
  Generative AI feeds off the data shared with it. So you have to be very careful about what information you offer to these platforms. Some companies, such as Samsung, have already learned this the hard way. Several employees used ChatGPT to check and optimize code, and transcribe meeting recordings into notes. Since these employees agreed to the terms and conditions of using ChatGPT, the AI absorbed this confidential information as part of its training.
  
  
• Intellectual property is a gray area
  What about copyrights or intellectual property rights for the contributions made by ChatGPT, and other authors? What happens if the platform plagiarizes a source, without giving it credit? Getty Images, for example, filed a complaint against Stable Diffusion because its AI platform trained on Getty Images’ content without permission.
  
  

ChatGPT and the other generative AI platforms hitting the market are proving to be immensely valuable for a variety of personal and professional purposes. They are a precious resource for speeding up routine tasks and automating processes. But their capabilities are not without limits.

What does the future have in store? It's still too early to predict with any certainty. Many big players in the tech industry have already signed an open letter calling on AI researchers to apply the brakes. This hiatus, they claim, would offer a window to fully grasp the potential impact AI could have on society, and the many ethical issues surrounding it. While others believe this technology will revolutionize how we live and work - for the better.

Whatever direction it goes in, the cat is now out of the bag. Artificial intelligence is here to stay and businesses should explore how they can reap the benefits.