The Age of Virtual Assistants

In the ever-evolving landscape of technology, virtual assistants are quickly becoming indispensable. What started as a novel feature has transformed into a pivotal tool in both personal and professional settings. By 2028, virtual professionals are expected to comprise half of the US workforce, according to Market.us. This staggering prediction highlights the increasing reliance on artificial intelligence (AI) and virtual assistants in reshaping modern life.
 
The journey of virtual assistants began long before they became embedded in our smartphones and computers. One of the earliest instances of such technology dates back to the 1960s with “ELIZA,” a computer programme developed at MIT that simulated conversation by using pattern matching and substitution methodologies. However, it was Apple that truly brought virtual assistants into the mainstream with the introduction of Siri on the iPhone 4S in 2011. Siri marked the dawn of the modern digital virtual assistant era, sparking a wave of similar technologies such as Amazon’s Alexa and Google Assistant.

Voice-based assistants like Siri, Alexa, and Google Assistant are set to dominate the future. Research from PwC indicates that voice assistants will soon play a central role in our daily interactions with technology, streamlining everything from personal tasks to professional workflows.

The Role of AI in Shaping Virtual Assistants

Artificial intelligence has been the primary driver of virtual assistants’ development, and recent innovations promise even greater capabilities. Apple’s latest AI and machine learning advancements, notably through its new intelligence technologies, will revolutionise how virtual assistants function. These tools will enable virtual assistants to anticipate user needs more accurately and efficiently, further blurring the line between human and machine interaction.

AI, in particular, is positioned to enhance customer satisfaction dramatically. A recent study by the IBM Institute for Business Value, surveying nearly 20,000 global consumers, reveals that 55% of respondents are eager for virtual assistants to improve their shopping experiences. Moreover, 59% expressed interest in AI applications that provide personalised information and services. This shows a clear trend towards AI integration in customer service, where virtual assistants will assist, predict, and fulfil consumer needs.

With this growth, however, comes a range of ethical concerns. Greg Duckworth, Principal AI/ML Consultant at Daemon, highlights the importance of ethical frameworks when implementing AI-driven virtual assistants. “This must be done with a strong ethical framework, especially in decision-making processes,” he emphasises. One of the critical issues is bias. Virtual assistants are only as unbiased as the data on which they are trained. If this data reflects societal biases, virtual assistants can perpetuate or even exacerbate these issues, particularly in hiring and customer service areas.”

Transparency is another pressing issue. Duckworth explains, “There is an ethical imperative to improve the explainability of AI systems so that decisions can be understood and scrutinised when necessary.” Without proper oversight, AI-driven decisions could lead to accountability problems, and errors might go unchecked.

Privacy and Data Security

Another significant concern is privacy. As virtual assistants become more integral to our lives, they inevitably collect vast amounts of personal data. Carsten Eriksen, Founder and CEO of Swift Creatives, addresses this issue: “Apple has made privacy a cornerstone of its ecosystem… data is processed on-device as much as possible, minimising the need to send information to external servers.” In contrast, other tech giants, such as Google, rely heavily on user data for targeted advertising, raising concerns about how much control consumers have over their personal information.

Eriksen warns that, as virtual assistants evolve, the debate between privacy and convenience will intensify. “Users need to be aware of how their data is handled and choose platforms that align with their privacy preferences,” he says. This debate will likely become more significant as virtual assistants become increasingly embedded in our daily routines.

The Future of Virtual Assistants

Virtual assistants are not just a passing trend. According to Jim Dvorkin, Senior Vice President of CX Products at RingCentral, they are set to reshape our professional lives over the next decade. “Virtual assistants are evolving beyond simple task management to become proactive, context-aware companions that can anticipate and meet user needs in real-time,” Dvorkin states. This transformation is particularly evident in customer-facing roles, such as contact centres, where virtual assistants enhance agent-customer interactions and streamline workflows.

The integration of generative AI and data analytics is key to these advancements. Virtual assistants are becoming increasingly adept at processing vast amounts of information, offering personalised interactions, and continuously learning from previous experiences. This not only improves customer satisfaction but also enhances business efficiency. Gopi Polavarapu, Chief Solutions Officer at Kore.ai, notes that virtual assistants offer a personalised shopping experience by “monitoring and processing buyer’s behavioural data and accurately predicting their next behaviour.”

As AI continues to evolve, so too will virtual assistants. They are no longer limited to basic tasks but are becoming integral to business operations and personal productivity. The ethical challenges surrounding bias, transparency, and privacy must be carefully addressed to ensure these technologies benefit society as a whole. As Carsten Eriksen aptly puts it, the future relationship between consumers and virtual assistants “will likely evolve into a partnership based on trust, personalisation, and shared goals.”

Virtual assistants represent not just a technological tool, but a transformative force poised to redefine how we live and work. The age of virtual assistants is not just on the horizon—it has already begun.

Perry Carpenter, Chief Human Risk Management Strategist at KnowBe4.

How do you foresee the role of virtual assistants evolving in both personal and professional environments over the next decade?

“In the next decade, virtual assistants are going to be far more complex and nuanced than many people realize. Yeah, there will be very predictable trends advancements like AI managing our schedules, handling emails, and optimizing our smart homes, helping with research, etc. But the changes that’ll catch us off guard are likely to be more substantial… and potentially concerning.

“As virtual assistants become more sophisticated, significant portions of the population will undoubtedly begin forming emotional bonds with them. We’ve already seen that in reports from OpenAI about the initial testing of their Advanced Voice Mode. Here’s a short passage from their GPT-4o System Card that they published on Aug 8:

During early testing, including red teaming and internal user testing, we observed users using language that might indicate forming connections with the model. For example, this includes language expressing shared bonds, such as “This is our last day together.” While these instances appear benign, they signal a need for continued investigation into how these effects might manifest over longer periods of time.  More diverse user populations, with more varied needs and desires from the model, in addition to independent academic and internal studies will help us more concretely define this risk area.

“People are already using virtual assistants as ways to provide basic (and even intimate) companionship. Now imagine an AI that has deep knowledge about your moods, your instincts, your motivations, your deepest fears, hopes, and dreams… and with the ability to know these even better than your closest friends (or you, yourself). This level of intimacy with an artificial entity raises profound questions about the nature of relationships, emotional dependency, and how that can be used in both positive and negative ways.

“In the professional realm we’re looking at a double-edged sword. Virtual assistants could revolutionize decision-making processes and provide insights and analysis at a speed, depth, and scale impossible for any human. But this can might cause us to over-rely on these AI advisors. We may miss mistakes, blindly follow data inaccuracies and data biases, or not realize that the output we receive was based on us asking the wrong questions. There’s also a real risk of human skills atrophying as we outsource more of our cognitive tasks to AI.

“There are also less intuitive use cases and issues. We will likely see virtual assistants taking on roles we never expected. For instance, AI mediators in legal disputes… offering the promise impartial, data-driven resolution suggestions. We’ll also see AI therapists, providing 24/7 mental health support. These applications sound beneficial, but they also raise ethical questions about the limits of AI involvement in deeply human issues. These uses also forget about potential fundamental biases that exist in the training data and how those biases might negatively impact the assistant’s recommends and interactions.

“The recent acqui-hire of Character.AI’s team by Google underscores the industry’s recognition of the potential here. The rapid growth of platforms like Character.AI demonstrates the increasing integration of AI assistants into our daily lives (interesting stats about Character.AI here. However, this popularity also raises concerns about data privacy and the potential for these systems to be used for disinformation.

“With the increasing popularity of services like Character.AI, it’s clear that people see value in interacting with AI personas in deeply meaningful ways. This level of engagement is unprecedented and hints at a future where the line between digital assistant and digital companion becomes increasingly blurred.

Here’s the way I tend to talk about issues like this: We’re entering what I call the “exploitation zone” – a period where technology is advancing faster than most of society’s ability to use the technology or fully grasp its implications (including potential misuses). It’s like we’re building a high-speed rail system while we’re still getting used to bicycles. The potential for both innovation and weaponization is enormous.

As we move forward, we need to be incredibly thoughtful about how we integrate AI assistants into our lives. We need to establish clear ethical and practical guardrails. This isn’t just about what these AIs can do for us, but how they might change us – our behaviours, our relationships, our very way of thinking. The challenge of the next decade won’t just be technological — it’ll deeply impact how we think, how we relate to others, and how we relate to ourselves.”

What are the key technological advancements driving the development of more sophisticated virtual assistants today, especially in retail?

“The tech behind today’s virtual assistants is evolving extremely rapidly. We’re seeing advancements that would have seemed like magic just a few years ago.

“The advent and popularization of large language models (LLMs, such as OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini, and Meta’s Llama) serve as linguistic powerhouses, allowing virtual assistants to engage in natural, context-aware conversations. It’s like they’ve gone from speaking a Siri-ish flavour of “caveman” to suddenly becoming virtual Shakespeare.

“AI models are also becoming multi-modal. Many now have the ability to process text, images, and audio rather than being forced to specialize in just one. In retail, this translates to virtual assistants that can not only tell you about a product, but also intelligently and eloquently tell you how the product fits into your life… moreover, the ‘assistant’ will understand enough about us to know exactly how best to manipulate us to purchase a specific product or take a certain course of action.

“On the positive side, we can easily imagine a virtual assistant that knows your style better than your best friend and can predict what you’ll want before you even know you want it. Or an AI that can optimize a global supply chain in real-time, making sure the right product is in the right place at the right time.

“These advancements are thrilling, but they also come with responsibilities. It’s like giving a teenager the keys to a Ferrari – amazing potential, but you want to make sure they understand the power they’re wielding.”

In what ways can virtual assistants enhance productivity and efficiency for businesses across various retail sectors?

“In a lot of ways, you can think of virtual assistants in retail as like having a super-powered employee who never sleeps, never takes a coffee break, and can be in a thousand places at once.  

“Think about customer service. AI assistants can handle a torrent of customer queries 24/7, freeing up human staff to deal with the most complex issues that require that special human insight or the kind of care/empathy that only a human can (currently) provide.

“In inventory management, virtual assistants can help predict demand, optimize stock levels, and streamline the supply chain with uncanny accuracy. Imagine never running out of that must-have item during the holiday rush – that’s the kind of superpower we’re talking about. And it’s important here to realize that these kinds of capabilities have been evolving steadily for several years now through traditional machine learning prediction systems and automation– not just after the advent of generative AI. However, generative AI adds another layer that has the possibility of taking these to a new level by adding even more natural language-based communication capabilities into the backend systems.

“Data analysis is another area where virtual assistants can shine. They can crunch numbers faster than you can say “quarterly report,” providing insights that can shape business strategy. It’s like having a team of genius analysts working around the clock, spotting trends and opportunities that might slip past the human eye.

“But all of this comes with a warning and a reality check – and it’s a big one – we need to implement these systems carefully. A virtual assistant that messes up inventory or gives customers bad info creates enormous potential for damage. Current large language models can make up information (this is often called ‘hallucination’). In these cases, we’ve seen them do everything from offer deep price discounts to customers who shouldn’t receive them to “Rick Rolling” a customer who asks about training videos that don’t exist.

“With today’s systems, it’s critical to have a human-in-the-loop at some level. It’s all about finding that sweet spot between AI capability and human oversight. And – of course – when AI chatbots offer the promise of an always-on, infinitely scalable workforce, then we worry about the potential ripple effects on the job market itself.”

How do you address concerns regarding privacy and data security when consumers are using virtual assistants?

“Privacy and data security in the world of virtual assistants will require constant vigilance and some clever strategies. And it will require us to think about the unintended or malicious ways our data might be accessed and used/misused.

“Education is key. We need to help consumers understand what data these AI assistants are collecting, how it’s being used, and how to protect their digital footprint. And here’s a crucial point that often gets overlooked: many companies offering these virtual assistants are in it for your data. They’re like digital vampires, thirsting for your personal information. Users need to understand the business model behind their AI assistants. Are they really getting a free service, or are they paying with their data?

“Ultimately, it’s about empowering users to make informed decisions. We want people to enjoy the benefits of virtual assistants without feeling like they’re living in a digital fishbowl.”

What are the ethical implications of increasing reliance on AI-driven virtual assistants, particularly in decision-making processes?

“When it comes to the ethical implications of AI-driven decision-making, we’re entering largely uncharted territory. The map is still being drawn.

“One of the biggest concerns is biased data. AI systems can inherit and amplify the biases present in their training data. And – with many systems – a core part of the training data is the internet. And the fullness of the internet is (at the very least) a mixed-bag when it comes to data quality. There’s a mix of good data, bad data, hate-speech, parody, opinion… you name it. That’s why post-training fine-tuning and reinforcement learning are critical. But this isn’t fully reliable or predictable yet in the ways that most people might expect.

“There are also the issues around accountability. When an AI makes a decision, who’s responsible for the outcome? It’s not like we can put a virtual assistant in the witness stand. This becomes especially thorny when we’re talking about high-stakes decisions in areas like healthcare or finance (again, highlighting the current need for a human-in-the-loop).

“Job displacement is another ethical minefield. As virtual assistants become more capable, what happens to the human workers they replace? We need to think carefully about how we manage this transition to ensure we’re not creating a digital elite and leaving everyone else behind.

“In its current form, AI can be incredibly powerful. But it shouldn’t be the be-all and end-all. It’s like having a really smart advisor – you listen to their input, but ultimately, you make the call. The goal should be to use AI to enhance human decision-making, not replace it entirely. We want to create a future where AI and humans are partners, not competitors.”

What are the current limitations of virtual assistants, and how are researchers working to overcome these challenges?

“Virtual assistants today are a bit like teenagers – incredibly capable in some ways, but still prone to some facepalm-worthy mistakes. These limitations are like growing pains on the path to maturity. Cringe-worthy examples from Google and Microsoft come to mind.

“One of the biggest hurdles is context and nuance. Virtual assistants can struggle with the subtle layers of human communication – sarcasm, idioms, cultural references. They’re constantly playing a game of “Spot the Subtext,” and sometimes they miss spectacularly.

“One of the most concerning current limitations is what we call “hallucinations” – when AI generates false but plausible-sounding information. It can be very similar to when you ask a genius a question that they don’t know the answer to. When put on the spot, the might make up whatever answer seems most plausible to them at the time rather than admitting they don’t know. This can be anything from minor inaccuracies to complete fabrications, and it’s a major challenge for building trust in these systems.

“Researchers are tackling these issues head-on. They’re developing more advanced natural language processing techniques to help AIs better understand context and nuance. They’re working on emotional intelligence algorithms to help virtual assistants read between the lines of human communication.

“There’s also a big push to improve the training data and methods. It’s like sending these AIs to a more diverse, well-rounded school, hoping they’ll come out with a broader, more nuanced understanding of the world. But make no mistake about it, I believe biased data or biased training is something we will never fully overcome. AI systems will always be amplifying or suppressing some group’s view of the world at the expense of another group’s.

“As the tech evolves, the goal is to create virtual assistants that are more adaptable, more accurate, and better at truly understanding and responding to human needs. We’re definitely not there yet, rapid progress is being made.”

How can virtual assistants be designed to better understand and respond to diverse cultural contexts and languages?

“This comes back to that fundamental data bias problem I’ve mentioned a couple times now. The first step is diversifying the training data. We need to feed these AIs a smorgasbord of cultural perspectives, not just a Western-centric diet. It’s about ensuring that the AI has “tasted” a wide range of cultural “flavours,” so to speak.

“Some developers are taking this a step further by building in cultural sensitivity modules. It’s like giving the AI a crash course in global etiquette and cultural norms. In the future, virtual assistants may be able to smoothly switch between cultural contexts, like a digital chameleon.

“But again, all of this raises fundamental bias questions and concerns. With so many diverse worldviews that need to be served by digital assistants, there will be different versions of ‘truth’ that get amplified or suppressed based on who is using the system. Different regions of the world have different cultural norms, ethical / moral systems, views on history/philosophy/etc. So, what is the ‘truth’ that an AI system should reflect in its outputs?

 “This isn’t a computer science problem… this is a big-hairy societal issue that can’t simply be ‘solved’ in the code. The implications here need to be deeply understood.

“This isn’t just about creating more effective virtual assistants – it’s about ensuring that as AI becomes more integrated into our lives, it serves all of humanity, not just a privileged subset. We’re trying to build a digital world that’s as diverse and inclusive as our physical one. It’s a challenging task, but an essential one.”

How do you envision the future relationship between consumers and virtual assistants as these technologies become more integrated into our daily lives?

“I see a world where virtual assistants become deeply woven into the fabric of our daily lives. They will be personal advisors, health monitors, and even digital companions. Imagine having an AI that knows your habits, preferences, and health history better than you do yourself. It could remind you to take your meds, suggest healthier meal choices, or even detect early signs of illness.

“In the professional sphere, these assistants could become indispensable colleagues, handling everything from scheduling to data analysis to first drafts of reports. They might even sit in on meetings, taking notes and suggesting action items.

“In the online dating world, we’ve even hearing about companies looking to create “digital twins” for their subscribers so that each member’s virtual assistants can go on a first date together to evaluate compatibility.

“In all of this, it’s crucial that we maintain our human connections and our ability to think for ourselves. Recent research has shown that AI-driven persuasion, especially when enhanced by personalization, can significantly outperform human persuasion. A study found that debating with GPT-4 using personalization resulted in an 81.7% increase in the odds of higher agreement with opponents compared to debating with humans. This finding is both fascinating and alarming. It underscores the need for users to maintain a healthy skepticism, to always consider the source and potential biases of the information they’re receiving, even when it comes from a seemingly helpful and personable virtual assistant. In all things, we’ll need to learn to question the potential underlying motives/drivers of the companies who make the virtual assistants.

“The ideal future isn’t one where AI replaces human intelligence, but one where the two complement each other. We want virtual assistants that enhance our capabilities, not diminish them. It’s about striking a balance. The rise of AI and virtual assistants is like watching the dawn of a new era – it’s breathtaking, a little scary, and it’s going to change everything. 

“The need for digital literacy is absolutely huge… it’s one of the most important life-skills we need to promote. Being able to critically evaluate information is going to be as important as reading and writing. We need to teach everyone how to navigate this AI-enhanced landscape, how to spot misinformation, and how to use these tools effectively. We need to understand the strengths and limitations of the tech so we can make informed decisions about how, when, and where we use it.

“We also need to stay vigilant about the potential for AI to be used to create and spread disinformation. The technology for creating deepfakes and other synthetic media is advancing at a breakneck pace. We now live in a world where a motivated party can easily spin up multiple targeted narratives and flood the internet with whatever version of ‘reality’ they want to propagate that day, drowning out everything else.

“There’s also an urgent need for ongoing public discourse and policy development around AI ethics and governance. We’re writing the rules as we go, and we need to make sure everyone has a voice in shaping these policies.

“The challenge ahead is to ensure we are able to harness the power of AI while still maintaining our humanity. It’s a balancing act, but one that could lead us to a future that’s more efficient, more inclusive, and maybe even more human than we could have imagined.”