Human:(AI):Human - the new paradigm of communication
What will it look like to live in a world with AI intermediaries?
I wrote an email to a friend the other day. I had bullet points based on a meeting we’d been in but they were generally pretty scrappy, so I used some AI to help me form this into a more coherent account of what had happened - this created a nice structured (sort of long) message with everything tied up in a bow. I later found out my friend used AI on their end to turn my email into summary bullet points…
This is going to be the new normal - whether we like it or not. As we move towards a world with ubiquitous access to generative AI and a new generation of native tools with this tech at its core, we can expect AI to become more of an intermediary in the ways those with access communicate with the world and each other. Although digital interfaces have for the last 20+ years been a membrane for a lot of interaction, we're now entering a more active paradigm:
human:human → human:(AI):human
There are many ways this will manifest, many of which likely won’t be clear right now - but as a starter:
Linguistic translation: Full access to technology will mean we can seamlessly speak in any global language and receive it in any other. This starts with scanning a foreign menu, builds up to using voice in multiple languages on a device, and develops until we're all using wearable devices seamlessly doing this in real-time.
Tonal translation: as we have new access in our apps and interfaces, AI will become an active filter for our voice, adjusting our tone or phrasing as we communicate. I’ve had many conversations with people who already use AI to do this - “I want to write in harsh ways and have the AI make it sound nicer” or “I sound too polite and I want AI to make things more straight to the point”. We’re already seeing how this could become the new normal in every-day products like Apple Intelligence email tone adjustment.
Summarise and un-summarise: people will write things with AI that will then be fed through AI on the other side to interpret it (much like my email exchange). Taking ideas and expanding them for us, or taking long text and getting the summarised version. This is already being rolled out natively in email apps.
Content generation: more of the content we consume that’s ‘written’ by other people will really be through an AI ghost writer doing the heavy lifting of turning ideas into output. Whether it’s a section or a whole document, much more will be written for us.
Content consumption: more and more people will use AI as a filter to the information they consume. This will be to i) consume more, ii) access content that may have been previously inaccessible (for example, it's too technical), and iii) act as a triage to decide what’s worth consuming in its full form. Already, many outlets are leaning into this, with AI summaries of news stories, blogs, and more now featuring on main pages.
Facilitating exchange: as AI agents become more common in digital spaces, we'll see a rise in moderator AIs facilitating online discussions - this will likely become most immediately prominent in professional settings.
Simulating exchange: we're already on the cusp of a new moment where much communication has first been 'tested' with a digital twin of the intended recipient - from a chatbot roleplaying a domain expert, all the way to simulating entire groups.
Agentic assistants: the most active form of this will be when someone doesn’t even have to pretend to interact with you - maybe they just have an agent who can do it on their behalf. From booking a restaurant to exchanging information with a colleague. We’re already seeing the rise of generative AI bots replacing customer service operators and much more.
Throughout our lives and various social spheres, this will be configured in a variety of ways. It won’t only be us having personal AIs to control as we see fit, but others with their very own unique set ups and settings - leaving much of the interpretation of our communication beyond our control. Of course, this doesn’t totally differ to today where we can do little to influence how someone might interpret what we say, think, or produce, but in a much more tangible way these preferences will be actively configured on every side (jointly controlled by the guidelines put in place by those building this technology). In this world, we will more than ever manifest our preferences and worldviews in the technology we design and use (which in turn shapes our preferences and world views).
Beyond the individual level, we’ll see changes in shared social spaces as we adopt communal AI systems: an AI agent moderating our workplace chat; an AI agent helping us navigate the information in our businesses, institutions, or social groups; or omnipresent AI in social media channels.
human:(AI):human
human:(AI1):(AI2):human
human:(personal AI):(shared AI space):(personal AI):human
human:(AI):human:(no AI):human
(ambient AI):human:(AI):human
and so on…
The impact of this transition will, like with most things, be a mix of good and bad (I've already written on one of the better meta trends I think this could be a part of). I’m sure readers will all have some immediate visceral reactions to this idea as reading.
As with all technology, its impact will wholly depend on how we decide to design and deploy it. One thing I do feel certain of is that this transition will bring in a lot of change. Exactly how this plays out is up to us.