What if we got it all wrong with the AI capabilities?

Alisa Kim, Phd
6 min readNov 10, 2024

Final Part 4 of “AI is Coming, Now What Do I Do?”

We constantly search for ways to do everyday tasks faster, better, and with fewer mistakes. But what if the solution isn’t to do these things better but to reimagine how we do them alltogether? There are several areas where humanity, howeveradvanced, just keeps on failing. Today, with AI’s capabilities rapidly evolving in all directions, we can shift how we communicate, learn, manage, and interact by reframing these tasks entirely. Here are four everyday areas that could be transformed by rethinking, not just optimizing.

1. Life navigation

🤺Challenge: Humans are often excellent at setting big goals but struggle with follow-through. Discipline and incremental change do not get us excited. Tracking progress toward these goals over months or years tends to fade as we prioritize short-term tasks.

👩‍🔬Reinvention: An AI “Life Navigator” could be a personal, long-term strategist. This tool wouldn’t just remind you of deadlines; it would help you course-correct and dynamically adjust your plans based on progress, new information, and changing priorities. But even more importantly, this AI Life Navigator can help you achieve excellence by diversifying the repetitive tasks, tricking your brain into thinking it’s something new and exciting. At the same time, it’s another way to achieve a chosen goal. Think about mixing up your fitness practice, offering a different way to practice a new language, and building neuroplasticity exercises into unexpected activities. Setting a 10-year career goal might look like an increasingly delusional exercise in a changing world, but the Life Navigator could regularly offer suggestions for skills to build, projects to pursue, or connections to nurture, all keeping you on course. It would reduce the cognitive load of staying organized over the long term and could even adapt to shifts in motivation. Observing us can provide interesting feedback for behavioral therapy, public speaking, self-presentation, habit-building, etc. To this, I would add “keeping ourselves happy” — too many people fulfill their serotonin needs through consumerism and social media simply because other approaches might not be handy enough.

🚧Limitations: It sounds like a very complex system; building one would be hard. But even more importantly, it would have to learn fast. How do we ensure our navigator is up to par with the recent shifts and not setting us on a generic course? AI might not be there yet, but even systems that human professionals firmly guide would offer a significant improvement. Finally, who would own such overwhelming amount of data about you? No company should be able to hold it, but maybe that’s where we will finally see AI and blockchain having a love child.

2.Info state synchronization

🤺Challenge: We have been perfecting verbalizing and visualizing our thoughts for

centuries, yet there often remains a barrier to ensuring two people in communication share a common understanding. As the number of people grows, information updating velocity increases, and the cost of alignment can become intractable. This has implications on all levels but is specifically relevant at an organizational level. Any business is a complex collection of processes, and many people have to agree on where they want to go, often with limited time and resources. Even in highly digitized tech companies, we spend endless hours aligning with team members, attending meetings, and sending detailed summaries to ensure everyone is on the same page — errors still occur.

👩‍🔬Reinvention: Imagine AI-powered tools that could analyze each team member’s work progress, identify the critical data points, and automatically create tailored messaging that the members of the team review for correctness, filling in only final details. These AI-generated updates would save time and reduce misunderstandings and potential communication bias between individuals. Instead of everyone manually catching up, the AI would autonomously track and share the relevant state of the project, ensuring each team member receives only the information they need when they need it. This tool would also provide a broader context to the organizational structure managers, solving the problem of top-to-bottom misalignment that notoriously many companies struggle with.

🚧Limitations: A slice of information permanently resides solely in one’s head, so an AI’s ability to fully replace human communication remains quite a challenge. Furthermore, the system would have to track every movement at work to compare the ongoing activity to an expected course of action, which might be considered invasive or obnoxious.

3. Mediation of interests

🤺Challenge: The productivity management software industry worldwide is expected to reach a projected revenue of US $119,690 billion by 2028. Safe to say there is sufficient economic interest — we addressed it from different angles, built machines, to-do lists, etc. Trust establishment is one factor that gets attention relatively infrequently (mainly by the crypto enthusiasts). Several excellent papers cover the relationships between trust and productivity, but here is a quick insight summary by Deloitte. Long story short — most countries can gain up to 2% extra per capita GDP growth if 50% of the population “trusted more people.”

👩‍🔬Reinvention: I would focus on identity and intent verification without going too deep into smart contracts. Imagine a future where scams and fraud (at least at an individual level) are no longer possible, where you have a mediator who can observe a given process or transaction and ensure the benefit of all parties participating. It can extend even further into arbitration and mediation (outside of the legal system for now). An “AI Mediator” could take an impartial, data-driven approach to facilitate discussions and guide decisions. Imagine a virtual mediator that could help de-escalate conflict by suggesting rational approaches and understanding each party’s concerns. In financial negotiations, it could simulate future scenarios based on various decision paths, helping users make calm, informed choices. By adding structured, logical input to emotional situations, this AI would help users avoid rash decisions.

🚧Limitations: What is a fact and what is a source of truth are still wobbly questions. Who and what defines the actual state of reality at a given point? We all exist in our perception bubbles, making agreeing potentially tricky. We would ultimately need to collectively agree on trusting the AI to define a reasonable reality — a significant challenge.

4. Streamlined state-citizen relationship

(The mayor of Berlin recently announced that this is under consideration for the city — quite exciting)

🤺Challenge: Bureaucratic processes often involve endless paperwork, repetitive data entry, and strict approval chains. This systems are famous for delays and add unnecessary overhead to almost any administrative task. This often results in high internal complexity, chaotic waiting states, and general confusion for citizens.

👩‍🔬The Reinvention: AI could autonomously handle entire document flows instead of simply digitizing forms or automating data entry. Imagine an “Intelligent Document Agent” that automatically understands document content fills out forms, cross-references necessary data, and flags items requiring approval based on predefined policies. It could even communicate directly with other departments or institutions for authorizations. By reducing human input, this approach would transform bureaucratic interactions into near-instantaneous, seamless experiences. Complex processes like tax filings, license renewals, or grant applications could become self-completing workflows where individuals provide a review for confirmation. In addition, this tool would be valuable for democratic states where staying informed is critical for civic participation.

🚧Limitations: The error cost could be very high, but if these error rates are low, the human apparatus could handle such cases very well. Bureaucratic processes are generally repeatable by the nature of legislation that defines them, so there is no good reason except the government’s mistrust.

Conclusion — let’s build

“The best minds of my generation are thinking about how to make people click ads. That sucks.” — Jeff Hammerbacher

I would love the conclusion to be — now let’s build. But I can’t help but recognise the growing amount of startups that put consumer benefits and general good first — and fail. Is it really impossible to survive without optimizing for profit? Or is it founders’ bias? I would love to dive deeper and understand this dynamics better, so if you have experience and thoughts on the matter, let’s have a discussion. At this point in time — I don’t see a reason to stop trying.

--

--

Alisa Kim, Phd
Alisa Kim, Phd

Written by Alisa Kim, Phd

Founder, Ex-AWS Bedrock, Stanford NLPLab alum, PhD in Machine Learning

No responses yet