August 2024

Artificial intelligence has revolutionized everything from customer service to content creation, giving us tools like ChatGPT and Google Gemini, which can generate human-like text or images with remarkable accuracy. But there’s a growing problem on the horizon that could undermine all of AI’s achievements—a phenomenon known as “model collapse.”

Model collapse, recently detailed in a Nature article by a team of researchers, is what happens when AI models are trained on data that includes content generated by earlier versions of themselves. Over time, this recursive process causes the models to drift further away from the original data distribution, losing the ability to accurately represent the world as it really is. Instead of improving, the AI starts to make mistakes that compound over generations, leading to outputs that are increasingly distorted and unreliable.

This isn’t just a technical issue for data scientists to worry about. If left unchecked, model collapse could have profound implications for businesses, technology, and our entire digital ecosystem.

What Exactly Is Model Collapse?

Let’s break it down. Most AI models, like GPT-4, are trained on vast amounts of data—much of it scraped from the internet. Initially, this data is generated by humans, reflecting the diversity and complexity of human language, behavior, and culture. The AI learns patterns from this data and uses it to generate new content, whether it’s writing an article, creating an image, or even generating code.

But what happens when the next generation of AI models is trained not just on human-generated data but also on data produced by earlier AI models? The result is a kind of echo chamber effect. The AI starts to “learn” from its own outputs, and because these outputs are never perfect, the model’s understanding of the world starts to degrade. It’s like making a copy of a copy of a copy—each version loses a bit of the original detail, and the end result is a blurry, less accurate representation of the world.

This degradation happens gradually, but it’s inevitable. The AI begins to lose the ability to generate content that reflects the true diversity of human experience. Instead, it starts producing content that is more uniform, less creative, and ultimately less useful.

Why Should We Care?

At first glance, model collapse might seem like a niche problem, something for AI researchers to worry about in their labs. But the implications are far-reaching. If AI models continue to train on AI-generated data, we could see a decline in the quality of everything from automated customer service to online content and even financial forecasting.

For businesses, this could mean that AI-driven tools become less reliable over time, leading to poor decision making, reduced customer satisfaction, and potentially costly errors. Imagine relying on an AI model to predict market trends, only to discover that it’s been trained on data that no longer accurately reflects real-world conditions. The consequences could be disastrous.

Moreover, model collapse could exacerbate issues of bias and inequality in AI. Low-probability events, which often involve marginalized groups or unique scenarios, are particularly vulnerable to being “forgotten” by AI models as they undergo collapse. This could lead to a future where AI is less capable of understanding and responding to the needs of diverse populations, further entrenching existing biases and inequalities.

The Challenge Of Human Data And The Rise Of AI-Generated Content

One of the primary solutions to preventing model collapse is ensuring that AI continues to be trained on high-quality, human-generated data. But this solution isn’t without its challenges. As AI becomes more prevalent, the content we encounter online is increasingly being generated by machines rather than humans. This creates a paradox: AI needs human data to function effectively, but the internet is becoming flooded with AI-generated content.

This situation makes it difficult to distinguish between human-generated and AI-generated content, complicating the task of curating pure human data for training future models. As more AI-generated content mimics human output convincingly, the risk of model collapse increases because the training data becomes contaminated with AI’s own projections, leading to a feedback loop of decreasing quality.

Moreover, using human data isn’t as simple as scraping content from the web. There are significant ethical and legal challenges involved. Who owns the data? Do individuals have rights over the content they create, and can they object to its use in training AI? These are pressing questions that need to be addressed as we navigate the future of AI development. The balance between leveraging human data and respecting individual rights is delicate, and failing to manage this balance could lead to significant legal and reputational risks for companies.

The First-Mover Advantage

Interestingly, the phenomenon of model collapse also highlights a critical concept in the world of AI: the first-mover advantage. The initial models that are trained on purely human-generated data are likely to be the most accurate and reliable. As subsequent models increasingly rely on AI-generated content for training, they will inevitably become less precise.

This creates a unique opportunity for businesses and organizations that are early adopters of AI technology. Those who invest in AI now, while the models are still trained primarily on human data, stand to benefit from the highest-quality outputs. They can build systems and make decisions based on AI that is still closely aligned with reality. However, as more and more AI-generated content floods the internet, future models will be at greater risk of collapse, and the advantages of using AI will diminish.

Preventing AI From Spiraling Into Irrelevance

So, what can be done to prevent model collapse and ensure that AI continues to be a powerful and reliable tool? The key lies in how we train our models.

First, it’s crucial to maintain access to high-quality, human-generated data. As tempting as it may be to rely on AI-generated content—after all, it’s cheaper and easier to obtain—we must resist the urge to cut corners. Ensuring that AI models continue to learn from diverse, authentic human experiences is essential to preserving their accuracy and relevance. However, this must be balanced with respect for the rights of individuals whose data is being used. Clear guidelines and ethical standards need to be established to navigate this complex terrain.

Second, the AI community needs greater transparency and collaboration. By sharing data sources, training methodologies, and the origins of content, AI developers can help prevent the inadvertent recycling of AI-generated data. This will require coordination and cooperation across industries, but it’s a necessary step if we want to maintain the integrity of our AI systems.

Finally, businesses and AI developers should consider integrating periodic “resets” into the training process. By regularly reintroducing models to fresh, human-generated data, we can help counteract the gradual drift that leads to model collapse. This approach won’t completely eliminate the risk, but it can slow down the process and keep AI models on track for longer.

The Road Ahead

AI has the potential to transform our world in ways we can barely imagine, but it’s not without its challenges. Model collapse is a stark reminder that, as powerful as these technologies are, they are still dependent on the quality of the data they’re trained on.

As we continue to integrate AI into every aspect of our lives, we must be vigilant about how we train and maintain these systems. By prioritizing high-quality data, fostering transparency, and being proactive in our approach, we can prevent AI from spiraling into irrelevance and ensure that it remains a valuable tool for the future.

Model collapse is a challenge, but it’s one that we can overcome with the right strategies and a commitment to keeping AI grounded in reality.

Discover the key differences between ChatGPT and generative AI. Learn how ChatGPT, fits into the broader category of generative AI and what sets it apart as a specialized tool for generating human-like content.

So, what exactly is the difference between ChatGPT and generative AI? Let’s break it down simply.

What Is Generative AI?

Generative AI is a broad term that refers to artificial intelligence systems specifically designed to create new content. This content can range from text and images to music, simulations, and videos. These AI systems leverage various models, such as large language models, Generative Adversarial Networks (GANs), or diffusion models, to produce outputs that are novel and original, learning from vast amounts of existing data.

Generative AI’s applications are extensive and diverse, impacting numerous industries and fields. For instance, in the creative arts, generative AI can produce music, design graphics, and even write poetry. In more technical fields, it can simulate complex processes or create detailed data visualizations. The underlying principle is the ability to generate new, meaningful content based on learned patterns and structures from existing datasets.

What Is ChatGPT?

ChatGPT is a prime example of generative AI, developed by OpenAI. It employs a type of machine learning model known as a large language model. ChatGPT excels in generating human-like text based on the prompts it receives. Initially celebrated for its conversational abilities, ChatGPT has expanded its functionalities to include generating images, writing computer code, and more.

The core technology behind ChatGPT, the Generative Pre-trained Transformer (GPT), allows it to understand and generate text in a way that mimics human language. This makes it incredibly useful for a wide range of applications, from customer service chatbots and virtual assistants to content creation and language translation.

However, the capabilities of ChatGPT have been expanded over the years to now also include the abilities to understand spoken words and speak back as well as understand and create images.

Scope and Specialization: The Key Differences

The main difference between ChatGPT and generative AI lies in their scope and specialization:

• Generative AI: This is the umbrella term that encompasses any AI system designed to create new content. Whether it’s writing articles, composing music, generating images, or simulating environments, all these fall under the generative AI category. Generative AI can use various models and techniques to achieve these creative tasks, making it a versatile and broad field.
• ChatGPT: A specific type of generative AI, ChatGPT is specialized in text generation. Think of it as a specialized tool within the larger AI toolkit, specifically sharpened for generating readable, coherent text. While it can now handle other tasks like generating images or writing code, its primary strength lies in its ability to engage in human-like conversations and produce natural language text.

Practical Implications

Understanding the distinction between ChatGPT and generative AI is important for grasping the broader capabilities and applications of AI technologies. While generative AI includes a wide range of systems and applications, ChatGPT exemplifies how these technologies can be refined and specialized for specific tasks.

For businesses and individuals looking to leverage AI, this means recognizing when a general generative AI system is needed versus a specialized tool like ChatGPT. For example, if the goal is to develop a virtual assistant capable of holding natural conversations, ChatGPT would be the ideal choice. On the other hand, for creating artistic designs or generating synthetic data, other generative AI models like GANs might be more appropriate.

To Sum Up

While ChatGPT is a powerful example of generative AI in action, it represents just one of the many applications of generative AI technologies in the world today. Generative AI is the broad category encompassing all AI systems that can create new content, while ChatGPT is a specialized tool within this category, focused primarily on text generation.

Understanding this difference helps you choose the right AI tool for your needs. Whether you’re a business owner, a student, or just curious about technology, knowing these distinctions can help you better navigate the world of AI.

Remember, the field of AI is constantly evolving, so keep learning and exploring these fascinating technologies.

The AI revolution is well underway, and I believe just about any business or organization can benefit by automating routine tasks, augmenting decision-making and optimizing operations and processes.

However, AI can also harm a business if it isn’t used cautiously. So, it’s very surprising to me that many companies and organizations don’t have any form of official AI policy in place.

Among the most serious risks are breaching privacy and confidentiality, exposing sensitive data, and inadvertently infringing copyright.

Creating such a policy should be at the top of just about every organization’s to-do list, regardless of size or industry. So, in this article, I’ll explore the risks that companies are exposing themselves to by allowing unregulated AI use, as well as the benefits of a well-thought-out policy when it comes to navigating the complex and sometimes dangerous waters of business AI.

Why Is Using AI Dangerous?

Long gone are the days when only large companies like Google or Microsoft were using AI. On a daily basis, millions of businesses are leveraging technology such as chatbots for customer support, generative content creation and audience analytics in marketing, screening job applicants in HR, detecting fraudulent transactions, optimizing supply chain operations or extracting business intelligence insights from their data.

Unfortunately, in my experience, many of them are unaware of the risks they’re leaving themselves open to.

Data privacy and security concerns are perhaps the most obvious, but still overlooked on a surprisingly frequent basis. Employees using tools like ChatGPT to create summaries or respond to emails are often unaware that they’re potentially exposing confidential information to the world.

Even if they are, some simply assume it isn’t a problem due to the fact they haven’t been told not to do it!

Several companies have already fallen foul of risks associated with a lack of regulation around AI.

For example, in 2023, Samsung banned the use of ChatGPT after finding that staff had entered sensitive data.

Another example is that HR departments routinely use AI tools to screen job applicants. However, unless proper care is taken to mitigate the risk of bias, this could lead to discrimination, potentially leaving the business open to legal action.

The same goes for businesses that are using AI tools that make decisions that can affect people’s lives – for example, processing loan applications or allocating healthcare resources.

When it comes to IP and copyright issues, businesses relying on AI-generated content could inadvertently find themselves using content without permission. Several court cases are currently being brought by artists and news agencies, saying their work was used to train algorithms without their permission. The outcome is uncertain right now, but could potentially lead to trouble further down the road for businesses using these tools.

And accountability is also an important issue. Are businesses and employees fully aware of their need to take responsibility for decisions that AI makes on their behalf? A lack of transparency and explainability inherent to many AI systems may make it difficult for them to do so. But this is unlikely to work as an excuse if they should find themselves in hot water due to their actions!

Getting any of this wrong could cause huge financial, legal and reputational damage to a company. So what can be done?

How An AI Policy Mitigates Risk

If a business wants to take advantage of the transformative opportunities offered by AI, a clear, detailed and comprehensive AI policy is essential.

Establishing guidelines around what constitutes acceptable and unacceptable use of AI should be the first step in safeguarding against its potential risks. However, it’s crucial to understand that an effective AI policy goes beyond mere risk mitigation – it’s also a powerful enabler for innovation and growth.

A well-crafted AI policy doesn’t just defend; it empowers. By clearly outlining how AI should be used to enhance productivity and drive innovation, it provides a framework within which employees can confidently explore and leverage AI technologies. This clarity fosters an environment where creative solutions are nurtured within safe and ethical boundaries.

Addressing these issues proactively will also help businesses identify the technological elements necessary for the safe and responsible use of AI.

For example, understanding the data policies around public cloud-based AI tools such as ChatGPT allows businesses to recognize where more private, secure systems — such as on-premises infrastructure – could be essential.

With this policy in place, any organization positions itself on far firmer ground. Rather than stifling them, it will empower organizations with the knowledge that they can experiment and innovate with confidence. An AI policy acts as a launchpad, setting up a framework for responsible and effective AI use that can drive competitive advantage.

The rapid adoption of AI across industries and the risks that this has created means an AI policy isn’t just a good idea — it’s critical to future-proofing any business.

Additionally, putting an acceptable AI use policy in place helps a company to position itself as a serious player in the AI game, rather than just another business jumping on the bandwagon. In an era where AI capabilities are rapidly becoming a benchmark for industry leadership, having a clear AI policy positions your company as a responsible, forward-thinking player. This can be incredibly attractive to investors, partners, and top talent who prioritize ethical standards and corporate responsibility.

It also helps to demonstrate to customers, investors and other stakeholders that an organization is committed to building trust and implementing AI in a transparent and ethical way.

This will be invaluable when it comes to hiring and retaining talent. People with the skills and experience needed to implement organizational AI systems are highly sought-after. Naturally, they’re attracted to companies that are able to demonstrate that they are serious and mature in their outlook and practices when it comes to AI.

This is something that I believe all leaders need to prioritize if they want to benefit from the opportunities offered by AI. A comprehensive AI policy doesn’t only defend; it also enables. It clarifies for all employees how AI should be used to enhance productivity and innovation, fostering an environment where creative solutions are nurtured within safe and ethical boundaries.

Following a turbulent few weeks in the stock market, there’s a lot of speculation that the AI “bubble” may be about to burst.

Many companies that are involved in AI have seen the value of their stock rocket dramatically over the past year or so. The most celebrated example is Nvidia, whose stock price has more than tripled since last summer. Others heavily invested in AI, such as Google and Microsoft, have also seen large increases.

Big leaps in stock price—particularly when they seem difficult to link to the generation of real value—are sometimes seen by analysts as an indicator that prices are due for a correction.

It’s hard to shake the feeling that we’ve seen all of this before. The hype and excitement around the newly emerged internet, which peaked in early 2000, was swiftly followed by the bursting of the “dot-com bubble.” Many companies went out of business, the economy experienced a significant downturn, and many jobs were lost.

So, is this a bubble? While it is said that AI could add around $15 trillion to the value of the global economy, recent earnings reports from the likes of Google and Tesla have been less than stellar, leading to the recent dips in share prices. At the same time, there are reports that the general public is becoming more distrustful of AI and that businesses are finding it difficult to make money from it.

Does this mean that the AI revolution—touted as holding the solution to problems as diverse as curing cancer and saving the environment—is about to come crashing down around our ears?

Personally, I don’t think so. But even if it does, in the long run, that might not even turn out to be a bad thing for AI adoption.

Is This 2000 All Over Again?

There are certainly striking similarities between what we are seeing today and the economic and market conditions that led up to the great dot-com bubble burst of 2000.

Both periods saw huge increases in stock prices and company valuations, seemingly driven by hopes and expectations for the future rather than tangible present-day results.

This has led to big stock indexes like the S&P 500, which heavily consists of tech stocks, mirroring the weighting of internet companies back at the turn of the century.

Just as was the case with the internet back then, all large companies are seeking to increase their valuations by leveraging their commitment to AI, with its promise of unlocking huge profits and growth further down the road.

However, there are also some very important differences. For a start, today’s AI leaders—companies like Microsoft, Nvidia, Alphabet and Meta—are already hugely profitable. They have proven business models and solid, reliable revenue streams that most likely won’t dry up even in the unlikely scenario that all of their AI plans fail.

This simply wasn’t the case for many of the hundreds of companies that failed when the dot-com bubble burst—most of which you’ve probably completely forgotten about by now.

With “all their eggs in one basket,” so to speak, they were just not set up to survive even a temporary dip in the market’s enthusiasm for the internet gold rush.

Things are very different today. Even if it turns out we’re not quite ready to make the leap to a fully automated, AI-driven society just yet, Google and Meta’s advertising services, Amazon’s shopping business and Tesla’s cars will keep those companies going until we are.

In addition, their huge user bases and vast amounts of proprietary data serve as protection against upstarts hoping to sneak in and launch competing AI services under their noses.

However, it’s important to note that even these tech giants aren’t immune to external pressures. The ongoing Google antitrust case, for instance, could have far-reaching implications not just for Google, but for other major players in the tech industry as well. Nvidia is already facing two separate antitrust probes from the U.S. Department of Justice, focusing on its acquisition of Run:ai and alleged anti-competitive practices in the AI chip market. These legal and regulatory challenges could potentially reshape the landscape for Big Tech’s AI ambitions.

It’s also worth mentioning that while the established tech companies have diversified revenue streams, there are newer players like OpenAI and Anthropic that are primarily focused on AI. Unlike many dot-com era startups, these AI-centric companies benefit from the support and partnerships of larger tech firms. Moreover, the AI startup landscape is less crowded than the dot-com boom, with fewer companies gaining significant traction. This more concentrated field of serious contenders might lead to a different outcome than we saw in the early 2000s.

So, Is It Safe For Me To Invest All My Money In AI Stocks?

First of all, nothing I say here should be taken as investment advice, as I am not a financial advisor by any means.

But here’s my opinion. It’s clear that the hugely inflated valuations we’ve seen in the past year or two are indicative of a bubble; a correction could very well be on the cards and may even be underway right now.

Despite this, however, I believe the mid and long-term prospects for AI are incredibly strong. It certainly has the potential to be just as—if not more—transformative than the emergence of the internet in the late 20th century eventually proved to be.

I also believe that part of the reason we haven’t seen many companies achieving real tangible value with AI just yet is that many leaders still aren’t prepared for the AI revolution. You can read more about my thoughts about that here.

As the dot-com bubble burst, the companies that didn’t survive were those that failed to think strategically about how the internet would impact their business and the opportunities it created.

Meanwhile, those that approached the internet not just as an opportunity to bring an existing business model online but as a way to radically rethink it from the ground up —Amazon, Google, and Netflix are just three of the most obvious examples—not only recovered but reached even greater heights.

This is why, even for a firm believer in the potential of AI like myself, acknowledging the likelihood that we are in a bubble of some sort and the fact that there could be some turbulence ahead is important.

Market corrections serve to remind us that the true, transformative potential of revolutionary advances like AI (or, indeed, the internet) lies in their practical applications and the real value they create rather than in speculative excitement and hype.

By refocusing our attention on genuine innovation and its tangible benefits, we become better equipped to recognize this value and make more informed decisions about our investments in the future of AI.

credit: Bernard Marr

I recently compiled a list of some of my favorite generative AI tools for boosting workplace productivity. But what about life outside of work?

Is generative AI also useful for tracking, organizing, managing and automating those day-to-day tasks that we often find ourselves trying to fit into our busy schedules?

Absolutely. Some clever people have turned their hands to creating AI tools and applications for a ton of every day (and some not-quite-so-everyday) activities, from cooking and cleaning to keeping fit and managing finances.

Here’s an overview of some of those that could be particularly helpful in helping us save time on daily chores and activities, which can then be spent doing things that are more exciting and rewarding.

Five Of The Best Generative AI Tools For Everyday Life

Let’s first look at five outstanding generative AI tools that can significantly improve everyday life, ranging from managing finances and planning meals to optimizing travel. These innovative applications are designed not only to simplify routine tasks but also to enrich your daily experiences, freeing up time for more enjoyable pursuits.

Cleo

Cleo is a personal budgeting app that uses a conversational, generative interface to help users keep track of their finances and spending. It provides insights into spending habits in natural language, which users can query and interact with through the chatbot interface. Designed to be flexible and user-friendly, it offers a “roast mode” that will tell you off for frivolous spending, as well as a “hype mode” that uses motivating language to encourage you to save and invest.

Expedia

Travel search engine Expedia has built generative AI chat functionality into its iOS app, enabling it to act as a virtual travel agent. So rather than having to sift through 1.6 quadrillion combinations of dates, destinations and airlines available via its website, users can get simple answers to holiday queries in straightforward, natural language.

Expedia Group was one of the first major online services to partner with ChatGPT creator OpenAI, and its personalized travel assistant Romie is powered by GPT-4. As well as natural language searching, it lets users create personalized travel itineraries and summarize guest reviews for quicker reading.

ChefGPT

Who doesn’t sometimes get stuck when it comes to thinking of ideas for dinner? Or find themselves looking at a kitchen stocked with random ingredients that don’t seem to go together?

ChefGPT aims to solve these everyday problems by acting as your personal chef and kitchen assistant. ChatGPT is comfortable with diverse diets, such as vegan or keto, and also works as a personal nutritionist, capable of helping users hit specific nutrition goals while keeping within their personal dietary restrictions. It even suggests perfect wine pairings for the meals that it generates.

Remodel

Home improvement and DIY are fun, but visualizing the result of your latest grand design can often be tricky. This app lets you create realistic visualizations for any interior or exterior design or landscaping project you can think of. Simply upload an image or input room dimensions and instantly see how it could look when remodeled in a variety of styles.

FitGPT

AI-powered fitness coach that generates personalized workout schedules, meal plans and motivation. There are several generative AI fitness coach apps available today, but FitGPT stands out from the crowd thanks to its ability to tailor fitness programs around individual user data and its user-friendly approach to helping you hit your personal goals. It also offers dietary and nutrition advice and is set up to help anyone, no matter what their starting level of fitness is. It’s also totally free for users of the basic plan.

More Great Everyday Generative AI Tools

While the tools mentioned above have already demonstrated the power of generative AI in everyday tasks, there are even more innovative solutions available that can further enhance your daily routines. Here’s a look at additional generative AI tools designed to streamline various aspects of everyday life, making those routine chores a bit more manageable and freeing up more time for the things you enjoy most.

Botatouille

Kitchen assistant app that helps you come up with meal ideas based on the ingredients you have handy or your mood.

Breda

Breda is a generative AI solution for parenting, which can generate activity ideas, provide advice on childcare issues, or just support you with encouragement and mindfulness advice when times get tough.

Dishgen

Another kitchen assistant app that generates original recipe ideas from ingredients.

DressX

Simplify choosing outfits with automated style inspiration.

Fitbod

Another AI-powered workout planner for creating ultra-personalized workout routines and tracking progress toward fitness goals.

HomeSynch

Create cleaning and home maintenance schedules with this ChatGPT-4o-powered chatbot app.

Magnifi

This app offers a conversational interface that aims to help you make better investment decisions by linking to online brokerage accounts and analyzing your portfolio and strategies.

Medito

A personalized meditation and mindfulness coach that invites you to share your thoughts, feelings and goals before creating bespoke meditation sessions.

Packpoint

Generate lists of items to pack for traveling based on where you will be going, what the weather will be like, and what activities you have planned.

Rescape.ai

Generate landscape gardening ideas from text prompts and bring them to life visually.

Skyscanner Savvy Search

Streamline your search for flights and holidays using Skyscanner’s travel planning chatbot.

Taskade Seasonal Cleaning To-Do List

Taskade is a platform that lets you build and share AI agents for a variety of jobs, and this is a ready-built generator for building and organizing home cleaning schedules.

Wally

A personal budgeting app that automates tracking income and expenditure and can chat with you through its built-in WallyGPT chatbot.

Woof

This is an AI-powered pet care app powered by Google Cloud that plans to offer real-time advice and feedback to pet owners.

credit: Bernard Marr

Since generative AI exploded onto the scene with the release of ChatGPT (still less than two years ago, unbelievably), we’ve seen that it has the potential to impact many jobs.

Perhaps one that will most obviously be affected is customer support. After all, chatbots are a flagship use case for generative AI, and the process of transitioning from human agents to automated systems began long before the emergence of language models (LLMs).

Does this mean everyone working in the field should be looking for new jobs, as the days of humans staffing customer support lines and helpdesks are numbered?

I don’t think so. A report by Harvard Business Review found that of 13 essential tasks involved in customer support and customer service, just four of them could be fully automated, while five could be augmented by AI to help humans work more effectively.

So, let’s explore the ways in which I believe the day-to-day work of customer support agents will be disrupted. I’ll also take a look at how professionals in the field can adapt to ensure they stay relevant in the AI-powered business landscape of the near future.

The Role Of Generative AI In Customer Support

Perhaps one of the most obvious applications – and certainly one we’re seeing enthusiastic adoption of – is chatbots. In the past, most of us will probably have experienced the frustration of dealing with slow, clumsy and far-from-intelligent voice recognition and automated customer support technology. Today, thanks to the application of chatbots built on LLMs, bots can have conversations that are close to being as dynamic and flexible as those of humans.

This means that we will increasingly see them used to deal with routine inquiries. However, they will also become capable of providing personalized and instant responses across many more in-depth and edge-case customer support situations. This will free human agents up to spend time on cases that require a human touch. This might be those needing case-specific knowledge not found in data the AI can access, multi-faceted problems or those that require input and collaboration from different departments.

Generative AI can also be used to build and maintain support knowledge bases. They can be continuously kept up-to-date with the latest developments in best practices so that human agents will always have access to the most current information and insights.

It’s also great at sentiment analytics. In a support context, this means it can quickly analyze large volumes of tickets or inquiries, categorizing them according to the sentiment of the customer. This could even take place in real-time, for example, by guiding human agents on how to respond during person-to-person interactions.

Generative AI can also be used to draft automated but personalized responses to email inquiries, making sure that messages carry a consistent tone while providing customers with advice relevant to their specific issues.

As we move forward, we can also expect to see even more advanced applications.

Predictive customer support will focus on solving customer issues before they are even raised. This could involve automating warnings, messages or prompts to install updates based on alerts from other AI agents working elsewhere in the business. For example, if a number of users are having difficulty accessing a service, then other users who are likely to want to use the service could be warned beforehand, enabling them to make alternative arrangements. Ultimately, this will reduce the chance of losing customers due to poor support experiences.

How Can Customer Support Professionals Stay Relevant?

I don’t believe that we will immediately see mass human redundancy across customer support roles. After all, people will always be required to cope with unexpected and unique challenges that always occur. I do, however, believe that professionals in the field who prepare themselves for the AI revolution will increase their chances of remaining useful and valued.

While AI will take some jobs, new jobs will emerge. There will be opportunities for customer support professionals who want to transition into new positions, such as AI trainers, who improve the quality of automated responses, and AI-human collaboration specialists, who will manage the integration of AI into human workforces.

This will involve staying up-to-date with the latest developments in workplace trends and AI technology, as well as adopting a habit of continuous learning and upskilling.

More value will also be placed on those who show themselves to be adept at human, soft skills that machines don’t yet have a good understanding of. These include emotional intelligence, empathy, and complex problem-solving – all core skills in customer support.

On the flip side, showing that you are capable of collaborating with AI and using it to augment your ability to carry out routine everyday tasks more quickly and efficiently will also make you valuable.

Overall, I believe that the secret to success is to learn to treat AI as both a tool and as a partner. Rather than attempting to compete with it in order to stay relevant, learn how and when it can be used to boost your own efficiency and productivity. And focus on developing human skills that AI can’t replicate when it comes to solving customer problems and improving customer experience.

credit: Bernard Marr

As artificial intelligence continues to reshape industries, understanding which jobs remain secure is crucial. While AI’s impact on the job market is undeniable, not all roles are equally vulnerable. Most jobs will be augmented to some extent, but those involving complex human emotions, advanced decision-making, and creative thinking are generally safer from complete automation. Here are some of the most AI-resistant careers:

Skilled Tradespeople

Skilled tradespeople, such as electricians and carpenters, possess hands-on expertise and problem-solving skills that AI cannot replicate. These roles require working in varied environments, adapting to unique challenges, and applying practical knowledge in real-time. The physical dexterity, spatial awareness, and on-the-spot decision-making needed in these trades make them resilient to automation. While AI can assist with tools and planning, the nuanced skills of a tradesperson remain irreplaceable.

Healthcare Professionals

Healthcare professionals, including doctors, nurses, and therapists, are essential roles that AI can only partially augment. AI can assist in diagnostics, patient monitoring, and even robotic surgeries. Still, the core of healthcare revolves around human empathy, interpersonal skills, and ethical judgment—qualities that AI cannot replicate. Understanding patient needs, providing compassionate care, and making critical ethical decisions ensure that healthcare professionals remain indispensable.

Strategic Decision-Makers

Executives and entrepreneurs occupy roles requiring high-level strategic decision-making and leadership capabilities. These roles involve understanding complex, multifaceted problems, weighing risks and benefits, and making decisions that steer organizations toward success. While AI can provide data-driven insights and assist in analysis, the nuanced understanding of business dynamics, stakeholder interests, and long-term vision are uniquely human attributes safeguarding these jobs from automation.

Creative Professions

Creative professionals, such as artists, writers, and designers, bring originality and a personal touch to their work that AI struggles to match. Generative AI tools can perform simple creative tasks and even assist in the creative process, but the essence of creativity involves inspiration, cultural subtleties, and deep emotional connections. The ability to convey human experiences, emotions, and unique perspectives ensures that creative professionals remain at the forefront of their fields, with AI serving as a supportive tool rather than a replacement.

Emergency Responders

Emergency responders, including firefighters, paramedics, and police officers, perform roles demanding quick thinking, human judgment, and physical presence. These jobs require the ability to assess unpredictable situations, make life-saving decisions, and provide immediate assistance. The complex and dynamic nature of emergencies is beyond the capabilities of current AI technologies. The physical and emotional demands placed on emergency responders highlight the irreplaceable value of human skills in these critical roles.

Embracing Uniquely Human Skills

The roles mentioned above emphasize human interaction, emotional intelligence, and creativity—qualities that AI can’t replicate yet. As AI continues to evolve, it’s essential to focus on developing and enhancing these uniquely human skills to stay ahead. By leveraging our innate abilities in empathy, strategic thinking, and creativity, we can ensure a harmonious coexistence with AI and secure our place in the future job market.

While AI will continue to augment and transform various aspects of work, the importance of human expertise, intuition, and emotional intelligence remains paramount. Embracing change, continuously learning, and adapting to new technologies will be key to thriving in an AI-enhanced world. By understanding the roles that are most resistant to automation, we can better prepare ourselves and future generations for a dynamic and ever-evolving job market.

So, what job is most safe from AI? It’s the ones that require the irreplaceable touch of human skills—those that make us uniquely human.

credit: Bernard Marr

Shadow AI, the unsanctioned use of AI tools in the workplace, is becoming a significant concern for businesses. This article explores the risks and opportunities of this trend, offering practical strategies for harnessing employee innovation while maintaining security and compliance.

In the ever-evolving landscape of business technology, a new phenomenon is emerging from the shadows: Shadow AI. Much like its predecessor, shadow IT, this trend is characterized by employees using unauthorized AI tools to boost their productivity and streamline their workflows. While this may sound alarming at first, it’s actually a sign of an innovative and forward-thinking workforce. The challenge for businesses lies in harnessing this creative energy without compromising security or compliance.

The Shadow AI Dilemma

Picture this: Tom, a marketing executive, discovers ChatGPT and starts using it to draft compelling email campaigns. Meanwhile, Sarah in finance is leveraging an AI-powered analytics tool to gain deeper insights into company spending patterns. Both are seeing impressive results, but neither tool has been vetted or approved by the IT department.

This scenario is playing out in companies across the globe. A recent study by Salesforce found that 49% of people have used generative AI, with over one-third using it daily. In the workplace, this translates to a significant number of employees bringing AI tools into their daily tasks without official sanction.

The Risks Of Unchecked AI Use

While the initiative shown by employees like Sarah and Tom is commendable, unchecked AI use can pose serious risks:

1. Data Security: When employees input company data into public AI tools, they may inadvertently expose sensitive information. Samsung learned this lesson the hard way in 2023 when they had to ban ChatGPT after staff entered confidential data.
2. Compliance Issues: AI tools that haven’t been vetted may not meet industry-specific compliance requirements, potentially leading to regulatory violations.
3. Inconsistent Output: Different AI tools can produce varying results, potentially undermining the uniformity of products, services, or customer experiences.
4. Ethical Concerns: Unchecked AI use in areas like HR could lead to biased decision-making, opening the company up to discrimination claims.

Turning Challenge Into Opportunity

Despite these risks, Shadow AI presents a unique opportunity for businesses to tap into their employees’ innovative spirit. Here’s how companies can navigate this new terrain:

1. Acknowledge and Embrace Innovation

Instead of shutting down unsanctioned AI use, recognize it as a sign that your team is actively seeking ways to work smarter. This innovative drive is a valuable asset in today’s competitive landscape.

2. Develop Clear AI Policies

Create comprehensive guidelines that outline which AI tools are approved for use, what kind of data can be input, and how to use these tools responsibly. Communicate these policies clearly and frequently to ensure all employees are on the same page.

3. Provide Training and Resources

Offer workshops, webinars, and e-learning modules to educate employees on the safe and effective use of AI tools. This not only mitigates risks but also empowers your workforce to leverage AI to its full potential.

4. Encourage Open Communication

Foster an environment where employees feel comfortable discussing their AI needs and discoveries. This open dialogue can help IT departments stay ahead of the curve and identify valuable tools that could be officially adopted.

5. Invest in Secure AI Solutions

Consider developing or investing in AI tools that meet your specific business needs and security requirements. This proactive approach can provide employees with powerful, vetted alternatives to public AI tools.

6. Regular Security Audits

Implement regular security checks to identify any unsanctioned AI use and address potential vulnerabilities before they become serious issues.

The Future Of AI In The Workplace

As AI continues to evolve and integrate into our daily work lives, the line between sanctioned and unsanctioned use will likely blur. Forward-thinking companies will need to stay agile, continuously updating their AI policies and practices to keep pace with technological advancements.

Harnessing The Power Of AI Safely

Shadow AI doesn’t have to be a threat. By acknowledging employee innovation, implementing clear policies, providing education, and fostering open communication, businesses can turn this challenge into a competitive advantage. The goal is not to stifle creativity but to channel it through secure, compliant pathways.

In this new era of AI-enhanced workplaces, the most successful companies will be those that find the sweet spot between innovation and security. They’ll empower their employees to leverage AI tools while maintaining robust safeguards against potential risks.

As we navigate this exciting frontier, remember: the shadow of AI doesn’t have to be ominous. With the right approach, it can illuminate new paths to productivity, creativity, and success in your organization.

credit: Bernard Marr

AI is going to change healthcare forever. In the near future, we can look forward to revolutionary new cures and treatments, personalized medicine, and a new generation of hospitals and facilities where super-smart robots take care of everything from cleaning to brain surgery.

At least – so we’re told. Now, I’m a believer in the power of AI to change the world for the better. But I can also plainly see that there’s a lot of hype around it. Technology companies stand to make trillions from selling it, and they all want us to believe their models and algorithms are the ones that will change the world.

In a world that’s quickly becoming flooded with AI washing, it’s critical that we learn how to cut through the hype and marketing bluster. So here I’ll take a look at how well one of the most frequent claims – that it will transform the fields of healthcare, medicine and wellbeing – stands up to scrutiny.

Where Are We Today?

So far, attempts to improve research and delivery of healthcare using AI have provided some encouraging success stories, as well as some clear examples of over-exaggerated expectations.

In drug and vaccine discovery, for example, it accelerated the discovery of mRNA vaccines for COVID-19. The same technology is now being used to create new protections against many other diseases. And generative AI built on similar technology to ChatGPT has also been used to create new immunotherapy drugs.

It has been used to analyze and interpret medical scans, images, medical records and tissue samples, and has been shown to be able to spot signs of diseases such as cancer and Alzheimer’s disease.

According to the American College Of Surgeons, “Most research shows that scan interpretation from AI is more robust and accurate than those from radiologists, often picking up small, rare spots in the images.”

It can shorten the length of time we need to stay in hospital, with one study of patients diagnosed with pulmonary embolism finding that those triaged with AI tools were able to return home an average of two days earlier.

The use of chatbots to answer patient questions or assist with the training of clinical staff, as well as analytics tools to help hospitals plan their use of resources more efficiently, all promise to streamline delivery and ultimately improve outcomes.

However, that isn’t to say that there haven’t also been missteps and over-inflated claims, particularly in the early days. In 2017, a partnership between IBM and MD Anderson Cancer Center was discontinued after it was found that the Watson Health system’s decision-making wasn’t accurate enough, creating a “credibility gap” with physicians.

In 2002, a study of online symptom-checker and self-diagnosis tools, some of which involve using AI to provide self-triage, found that “overall, the diagnostic and triage accuracy of symptom checkers are variable and of low accuracy … this study demonstrates that reliance upon symptom checkers could pose significant patient safety hazards.”

It should also be noted that despite the fact that many hospitals and healthcare providers are exploring the use of AI systems for creating efficiencies in administerial, record-keeping and staffing processes, it’s hard to find evidence that this has yet led to significant cost savings or improvements in patient care.

What Are The Experts Saying?

If we look beyond the words of marketers and the CEOs of big tech companies, opinions vary on the value that AI has – or will – bring to the field of healthcare.

Geneticist and cardiologist Dr Eric Topol, author of Deep Medicine, says that while AI will probably never replace a thorough physical examination by an experienced doctor, many applications are already proving highly valuable. These include self-administered kits that test for urinary tract infections and analyze lung problems from the sound of a cough.

In the future, he believes, mobile phones are likely to provide inexpensive ultrasound scans, and recognize early warning signs of diabetes from a picture of the retina.

Perhaps most valuable of all, he suggests, will be the ability of AI to restore the “human element” to medicine. By handling routine tasks, AI will free up doctors to spend more face-to-face time with patients, enabling them to develop a better understanding of their condition.

This is a position echoed by Dr Fei-Fei Li, director of Stanford University’s Institute for Human-Centered AI, who says “We’re seeing the medical profession using AI technology … [doctors] tell me that medical summaries are very painful … they take away from patients. Now you can get a language model to help.”

However an article in MIT Technology review presents a more cautionary outlook. Tools designed to diagnose disease or predict outbreaks could be trained on limited or biased data – for example, research has shown that some tools are less effective with female patients simply because less women take part in medical studies.

And Dr. Robert Glatter, assistant professor of emergency medicine at the Zucker School of Medicine, together with Dr. Peter Papadakos, professor of anesthesiology and surgery at the University of Rochester Medical Center, in their article No AI Can Ever Learn The Art Of Medicine, argue “Even with the advent and ongoing evolution of AI … empathy is best learned and communicated in the form of bedside teaching by humans – not AI or chatbots.”

The Verdict – Hype Or Reality?

While AI has undoubtedly already made significant impacts and seems likely to become more valuable as we move forward, it’s unlikely to be the literal panacea that will cure all of our ills.

However, we’re in the very early days of the AI revolution, and successes in fields such as drug discovery and detection of early warning signals clearly demonstrate that the potential to improve care and outcomes is real.

Challenges – particularly around data privacy, implementation costs and training of healthcare professionals in its use – will need to be overcome before the most optimistic predictions become viable.

As in other fields, professionals agree that its most valuable applications will involve augmenting rather than replacing human skills, experience and expertise.

For the technology companies building tools, keeping this firmly in mind while developing the next generation of AI-enhanced medical applications will be key to driving real change that will benefit us all.

Credit: Bernard Marr

he AI revolution is well underway, and I believe just about any business or organization can benefit by automating routine tasks, augmenting decision-making and optimizing operations and processes.

However, AI can also harm a business if it isn’t used cautiously. So, it’s very surprising to me that many companies and organizations don’t have any form of official AI policy in place.

Among the most serious risks are breaching privacy and confidentiality, exposing sensitive data, and inadvertently infringing copyright.

Creating such a policy should be at the top of just about every organization’s to-do list, regardless of size or industry. So, in this article, I’ll explore the risks that companies are exposing themselves to by allowing unregulated AI use, as well as the benefits of a well-thought-out policy when it comes to navigating the complex and sometimes dangerous waters of business AI.

Why Is Using AI Dangerous?

Long gone are the days when only large companies like Google or Microsoft were using AI. On a daily basis, millions of businesses are leveraging technology such as chatbots for customer support, generative content creation and audience analytics in marketing, screening job applicants in HR, detecting fraudulent transactions, optimizing supply chain operations or extracting business intelligence insights from their data.

Unfortunately, in my experience, many of them are unaware of the risks they’re leaving themselves open to.

Data privacy and security concerns are perhaps the most obvious, but still overlooked on a surprisingly frequent basis. Employees using tools like ChatGPT to create summaries or respond to emails are often unaware that they’re potentially exposing confidential information to the world.

Even if they are, some simply assume it isn’t a problem due to the fact they haven’t been told not to do it!

Several companies have already fallen foul of risks associated with a lack of regulation around AI.

For example, in 2023, Samsung banned the use of ChatGPT after finding that staff had entered sensitive data.

Another example is that HR departments routinely use AI tools to screen job applicants. However, unless proper care is taken to mitigate the risk of bias, this could lead to discrimination, potentially leaving the business open to legal action.

The same goes for businesses that are using AI tools that make decisions that can affect people’s lives – for example, processing loan applications or allocating healthcare resources.

When it comes to IP and copyright issues, businesses relying on AI-generated content could inadvertently find themselves using content without permission. Several court cases are currently being brought by artists and news agencies, saying their work was used to train algorithms without their permission. The outcome is uncertain right now, but could potentially lead to trouble further down the road for businesses using these tools.

And accountability is also an important issue. Are businesses and employees fully aware of their need to take responsibility for decisions that AI makes on their behalf? A lack of transparency and explainability inherent to many AI systems may make it difficult for them to do so. But this is unlikely to work as an excuse if they should find themselves in hot water due to their actions!

Getting any of this wrong could cause huge financial, legal and reputational damage to a company. So what can be done?

How An AI Policy Mitigates Risk

If a business wants to take advantage of the transformative opportunities offered by AI, a clear, detailed and comprehensive AI policy is essential.

Establishing guidelines around what constitutes acceptable and unacceptable use of AI should be the first step in safeguarding against its potential risks. However, it’s crucial to understand that an effective AI policy goes beyond mere risk mitigation – it’s also a powerful enabler for innovation and growth.

A well-crafted AI policy doesn’t just defend; it empowers. By clearly outlining how AI should be used to enhance productivity and drive innovation, it provides a framework within which employees can confidently explore and leverage AI technologies. This clarity fosters an environment where creative solutions are nurtured within safe and ethical boundaries.

Addressing these issues proactively will also help businesses identify the technological elements necessary for the safe and responsible use of AI.

For example, understanding the data policies around public cloud-based AI tools such as ChatGPT allows businesses to recognize where more private, secure systems — such as on-premises infrastructure – could be essential.

With this policy in place, any organization positions itself on far firmer ground. Rather than stifling them, it will empower organizations with the knowledge that they can experiment and innovate with confidence. An AI policy acts as a launchpad, setting up a framework for responsible and effective AI use that can drive competitive advantage.

The rapid adoption of AI across industries and the risks that this has created means an AI policy isn’t just a good idea — it’s critical to future-proofing any business.

Additionally, putting an acceptable AI use policy in place helps a company to position itself as a serious player in the AI game, rather than just another business jumping on the bandwagon. In an era where AI capabilities are rapidly becoming a benchmark for industry leadership, having a clear AI policy positions your company as a responsible, forward-thinking player. This can be incredibly attractive to investors, partners, and top talent who prioritize ethical standards and corporate responsibility.

It also helps to demonstrate to customers, investors and other stakeholders that an organization is committed to building trust and implementing AI in a transparent and ethical way.

This will be invaluable when it comes to hiring and retaining talent. People with the skills and experience needed to implement organizational AI systems are highly sought-after. Naturally, they’re attracted to companies that are able to demonstrate that they are serious and mature in their outlook and practices when it comes to AI.

This is something that I believe all leaders need to prioritize if they want to benefit from the opportunities offered by AI. A comprehensive AI policy doesn’t only defend; it also enables. It clarifies for all employees how AI should be used to enhance productivity and innovation, fostering an environment where creative solutions are nurtured within safe and ethical boundaries.

credit: Bernard Marr