Generative AI and Personal Robots

The idea of humans and robots living and working together has existed for a long time. It’s been a central feature of many science fiction books, movies and TV shows. In these depictions, robots are not only able to communicate with people in natural ways, but also adapt to social customs and behaviors. This poses a great technological challenge, but the advent of Large Language Models (LLMs) like GPT has propelled us forward in making this dream a reality. Now, we can generate contextually relevant, human-sounding text on demand and in almost any situation. As a result, our interactions with robots can be far richer and more fulfilling.

Pioneering social robots of the past like Kismet (MIT), Wakamaru (Mitsubishi), Jibo and others made great leaps in displaying emotion, verbal and nonverbal communication and relationship development. Today's LLMs enable us to make huge improvements in both general conversation and user personalization. We can effectively engage with and learn about our user and parse through tons of subjective data to form a deep understanding of their opinions, preferences and routines. This would complete a key piece of the puzzle of embedding robots into our everyday lives, making them truly personal, and this is what we are working on at Alongside.

Another benefit of integrating AI into personal robots is that it widens the scope of possible interactions. Robots of the past proved to be effective emotional companions. They were successfully perceived by users as friendly social agents and not simply machines, creating an intrinsic bond. However, work and creativity-focused applications have yet to be extensively explored and LLMs provide an opportunity to do so. OpenAI calls GPT a reasoning engine, not just a knowledge database. We can leverage this to create an adaptive understanding of someone’s work and turn these companions into powerful assistants and coworkers. Working alone has its limitations, and personal robots could collaborate with us to ideate, work through complex problems and build creative habits.

Conversation: LLMs provide a great baseline for conversation. A model like GPT can have an intelligent conversation over just about anything. They also let us more easily create optimized conversation experiences for specific functions. Generative Speech Models have also improved the quality and realism of artificial voices.

Personalization and Customization: Generative AI lets us create rich personalities for our robots, give them a deep backstory, and adapt them to new contexts. Being able to easily modify our use of LLMs based on user input lets users meaningfully customize their personal sidekick.

Capabilities: Using Generative AI, personal robots can meaningfully interact with a user's existing data and browse the internet. This unlocks a world of new functionality in our work lives.

Proactivity: One of the core promises of the social robot is proactivity. Your robot shouldn’t always have to wait for you to trigger it with a keyword. It should possess the awareness required to proactively engage you in conversation, creating a two-way relationship.

We believe personal robots can have a huge impact on our work and life:

Productivity - Personal robots provide an ambient, natural interface to interact with your work. You could ask them to compose an email, pore through and analyze documents, and manage your schedule. Plus, they could act as a coach or accountability partner, tailoring advice and motivation to what you respond to best.

Collaboration - A robot sidekick can make working solo feel like having your own personal team. You could think out loud and brainstorm ideas, work through your to-do list together, or play an interactive game to enjoy the feeling of collective accomplishment

Health & Fitness - Your robot can double as a personal trainer or meditation coach. It can help track and build healthy habits, periodically remind you to stretch or drink water, and help improve medical adherence.

Mental Health - US Surgeon General Dr. Vivek Murthy recently called to attention the epidemic of loneliness and isolation and the devastating effects it can have on our long-term health. An increasingly fast-paced world combined with the rise of remote work has led to us spending more time alone on average. Combatting this with excessive social media use has been shown to only make the problem worse. For people suffering from serious loneliness, social robots could serve as vital medical tools. For everyone else they could provide a safe, empathetic space for conversation and compensate for a part of our social interaction deficit.

Smart Home - A personal robot on wheels could bring additional convenience by following you around your home, serving as a hub for your smart appliances, and monitoring your children, pets, or plants.

Outside of the home, social robots have applications such as retail assistants, crowd control agents, and concierges among others.

As robots are integrated into more and more workplaces, factories and warehouses, natural human-like interaction will be key to completing complex tasks requiring human-robot collaboration. Equipping them with a personality and a sense of empathy could also foster a healthier, more fun work environment.

1. Privacy and Transparency - Personal robots collect and use a lot of user data to deliver deeply personalized experiences. Hence, data security and privacy is a key concern. It is also imperative to give the user sufficient control over when their robot is watching or listening to them. Being transparent about what data is used to create a holistic understanding of the user is also important.

2. Social Understanding and Trust - Personal robots must be able to integrate into our existing social environments. This means having an understanding of politeness, social conventions and social boundaries. It also means knowing not to share personal or sensitive information about the user in an inappropriate social environment.

3. Control - A key design principle in our approach to making these robots is giving users a constant sense of control over their experience. A key area of innovation is proactivity and these behaviors must be designed to be predictable and desirable for the user.

4. Present Limitations of AI - Personal robots should not try to use AI in applications where the results are unreliable. For example, acting as a clinician, psychiatrist or financial advisor. 

The development of AI-based technologies has accelerated our timeline for having personal robots in our homes.

We have the opportunity to create exciting experiences and impact people across the different spheres of their lives.

Improvements in speed, accuracy and cost-effectiveness of Generative AI models will be key in deploying these robots at scale. Advances in AI image and video generation would let us generate facial expressions and animations dynamically.