By Emil Bjerg, journalist and editor
From help with coding and finances to emotional support and personalized recommendations, are we all getting personal AI assistants – umm, companions – in the very near future?
From big companies like Meta and Microsoft to upcomers like PI and New Computer, AI companions are suddenly everywhere.
AI companions could emerge as a defining technology of our time, blending the convenience of digital assistance with a previously unimaginable level of personalization. This fusion has the potential to reshape our interactions with technology – for better or worse – paving the way for a future where AI companions are not just tools but helpful partners in our daily lives.
Analysts expect that many, if not most, of us will have personal AI companions in the very near future. But how are they different from the personal assistants and other types of AI we already know?
Why Is It a Companion and Not an Assistant?
Traditional personal assistants like Siri or Alexa, despite substantial investments and advancements, have often been limited to performing simple tasks or acting as voice-to-text intermediaries. While at times helpful, they have yet to fundamentally change how we interact with technology.
The new generation of AI, branding itself as “companions” rather than “assistants,” signals a step towards more personalized interactions. These AI companions are designed to understand and adapt to individual preferences, moods, and needs, making the experience far more personal than the one-size-fits-all approach of traditional assistants.
Unlike earlier AI assistants that offered static responses, AI companions are equipped with advanced machine learning capabilities. They learn from interactions, adapting to provide more relevant and context-aware answers, making them more like companions who understand you over time.
AI companions are designed to assist in various areas – from professional tasks like coding and financial planning to personal well-being, including emotional support and leisure recommendations. This versatility sets them apart from the more task-specific assistants of the past.
Companions: A New Generation of Generative AI
Generative AIs like ChatGPT and Bard have marked a significant step towards realizing the concept of an AI assistant: suddenly, an AI could create a cheap weekly meal plan or help draft tedious emails. Unlike an actual assistant, however, ChatGPT or Bard quickly forget interactions, giving a feeling of starting over – over and over again.
The AI companions offer an essential, additional value proposition: They remember various inputs like a human assistant or friend would. We’ll get back to what opportunities that feature opens for human-AI interaction and collaboration. First, let’s have a look at who’s creating the new AI companions.
Both Big Tech and Newcomers Contribute
Both Big Tech and newcomers are creating new offerings.
On the more colorful side of the spectrum, Meta has upped the budget and hired a number of famous people to be the face of their AI companions – from Snoop Dogg to Kylie Jenner. By using well-known faces as the interface for their AI companions, Meta is attempting to bridge the gap between users and technology.
Some have dubbed the move dystopian, others “a bit too shameless to feel all that revolutionary.” But by infusing AI companions with familiar faces, they could also break down barriers, making the technology more accessible and relatable. Some users will hopefully be positive, so the tech company can justify the budgets: top creators have been paid up to 5 million US Dollars to allow Meta to use their identities as very vivid avatars.
Microsoft, per usual, has a more sober, professional approach to its AI companion. With their Copilot – an “everyday AI companion” – they have introduced a service that can help with anything from coding in Github to productivity in Microsoft 365 to how you search in their engines. This companion learns from user interactions, offering personalized suggestions and support. The idea from Microsoft is to create an AI companion that isn’t just a tool for specific tasks but a holistic partner in various aspects of professional work. A recent review from PCMag calls Copilot a ‘terrific research assistant.’
But it’s the approach to personalized AI from the newcomers that perhaps feels the most revolutionary.
Pi – led by Mustafa Suleyman, former Deepmind – and New Computer – created by Apple designer Jason Yuan and others – might become household names in the near future. While New Computer’s AI companion is still in the testing phase, Fast Company’s Mark Wilson has been able to try a pre-alpha version. The interface, he reports, is much like any messenger app or ChatGPT’s mobile interface. The difference lies in the memory – the ability to ‘get to know’ its user.
Wilson reports that the chatbot’s memory of each chat allows it to give on-point advice. In his use case, that ranges from help with choosing a new type of handwriting to pick up to how to avoid perfectionism when learning new things.
The personalization in AI companions allows them to provide value when they aren’t being prompted. Wilson reports that New Computer’s pre-alpha can browse the web for valuable, tailored insights while the user is away from the keyboard.
Another interesting question on the subject of AI companions is if we’ll see new types of hardware introduced. The most promising candidate for that is currently Humane’s AI, which recently featured on TIME’s Best Innovations of 2023 list. Wanting to deliver a screen-free future, the founders behind Humane have created a pin that the user can “ask complex questions to make calls and send texts, all using just your voice.” The pin has a built-in camera that can provide information on visual input – e.g., an estimation of the number of calories in a dish. The pin also has a projector that allows it to display visual information, even on the palm of a hand.
A new personalized way of AI isn’t without its ethical dilemmas – we’ll return to those shortly. First, let’s have a look at how personalized AI companions can change the way we work and interact with AI.
How Might This Change the Way We Work and Interact with AI?
While a number of AI companions are becoming publicly available, it’ll most likely be a while before they reach their full potential. In most cases, the full maturation of these technologies will depend on real-world usage and the data and feedback it generates. In the long run, with the companion’s ability to remember and ‘get to know’ a user, the potential is vast. For individuals as well as companies.
For the coder being stuck with a complex problem, an AI companion could be able to provide solutions or new perspectives overnight.
For an R&D department, a companion, given the right data, can synthesize trends and events from the external world with internal data and, from there, suggest improvements or new products.
AI companions are likely to help many of us fill our knowledge gaps. We can Google what we know we don’t know. Given an incomprehensible amount of data, an AI companion can help us learn what we don’t know that we don’t know.
Mark Wilson believes AI companions “will prove to be a pretty radical superpower for many people. “ He foresees that they’ll be used as a “reminder engine, a search buddy, an efficiency expert, an infinite storehouse for emotional labor.”
But with the exciting potential comes a series of ethical questions.
Ethical Questions
A main concern centers around the data-intensive nature of something personalized to this level. The companions can only give customized information and advice with a lot of data – of a kind that borders between the personal and the private.
As AI is likely to move from the less personal data-requiring chatbots to companions, do we feel safe sharing this type of information with tech companies? And maybe, more interestingly, will the companion nature of the AI ‘trick’ us into sharing more information than we’re comfortable with, especially as a possible attachment between the user and AI grows?
Bias, like data concerns, is always a theme worth discussing when talking about personalized AI. While AI companions ideally can help fill knowledge gaps, ensuring that they offer diverse and unbiased perspectives is crucial. Their ability to broaden our understanding rather than reinforce preconceptions will be essential in determining their value – economically and ethically.
Observers even warn that the new degree of personalization can create emotional complications as interdependence – whether of a professional or personal character – will deepen. “What happens when there’s a software update or when the algorithm changes? The emotional turmoil can be very real for some users”, Jamie Bykov-Brett writes.
Personal AI wearables can become an ethical minefield as well. As AI companions become hardware and users can easily record and film events and conversations in public, how will this affect the dialogues around consent?
In a world where technological breakthroughs are launched at an unprecedented pace, we’re all likely to navigate the immense potential and new ethical challenges that AI companions bring about – very soon.