Weeknotes 313 - the triage of AI search
Hi, y’all!
Today is a special day. Let’s hope for the best outcome, and a peaceful aftermath. Platformer has you prepared what to expect in (online) media. Whatever the outcome is, it will be looked back upon. Will it be like the ancient Rome collapsing?
Last week I extended the time for the explorative research project on civic economies and protocol economies as residency within the Civic Interaction Design research group of Amsterdam University of Applied Sciences, and funded by the Center for Economic Transformation. I will probably share more on this in the coming month, but one of the goals is to map the current pioneers in the field, so let me know if you know people or companies I should not miss!
Next, I am very happy to see how TH/NGS 2024 is taking shape. The program is complete; I updated the website with the list of sessions and speakers and will add more details this week. Check it out, and I hope many of you visit in December. As readers of the newsletter, I can offer a 25% discount when using the code: TARGET IS THINGS (valid until the next weeknotes next week).
Triggered thought
The Generative AI news of the week is all about the AI search engine war that seems to be starting to happen. I had to think about that other big shift in searching the web; the Google moment, and wondered if we are at the type of paradigm shift in digesting the information and knowledge the internet (in the broadest sense) is offering.
Back in the day, the shift was from expert-driven selection to the wisdom of the crowd. Google added PageRank as a new filtering mechanic and a new way of trusting that it would find the right things that fit your real question. It was inspired by academic referencing but made accessible to the rest of us.
But with the inevitable role of the internet in finding information for everything, especially the e-commerce angle, the systems started biting their tail and slowly breaking down the trust. Google mixed too many paid advertisers into the mix, and the search engine optimization as marketing became too dominant.
And now we are entering AI search. Perplexity paved the way, breaking down some holy houses (a Dutch expression; check the translation at OpenAI search), and this week, OpenAI is daring to enter that playing field. Meta also announced that it will be building its version.
There will be a discussion about trust in the results, hallucination, how referencing works, and the functionality compared. But I think it's more interesting to look at the new paradigm shift here.
We can no longer rely on PageRank for manipulation by Google and professional search engine optimization companies. And 20 years later the world became even more complex, and the digital information as a source untameable. That is the fertile ground for AI search, where inferring and machine learning assist us in filtering the right answers. With the capabilities of real-time machine learning combined with the proven power of a natural conversational style, we can now execute these tools.
There are caveats, though. The question of how human and AI power will be combined is not played out yet. In this research paper, the human-AI partnership is valued in different categories. It seems that humans using AI are not the best at adding value to collaboration, and also the other way around. When the goal is a creative process, the collaboration of the two improves the results.
So, is search a creative result? That depends on what you are looking for. If you are still not sure about what your question is, more of a hunch than a specific need, that is a clear fit for AI search. But that is a different search than finding the cheapest offer for your next vacuum.
These differences should be integrated into the new search functionality. The first action the AI search performs is not a search but a triage on what type of search is intended and starts a conversation about that. Depending on the type of question, traditional search can be used, or new capabilities of generative AI are mixed into the results and formulation to help out.
The rise of AI search is inevitable, let alone that people using AI tools are probably already doing it. That makes the OpenAI’s move potentially smart; try to filter the questions based on expectations and choose the type of ‘tool’ to provide the answer.
One question remains open: how will the commercial search results become integrated? PageRank made Google big and a dominant player, and then commercial interests started breaking down relevance. To prevent this with AI search, we should start developing a new user experience paradigm to have the offerings found by the user in an honest conversation. Can the AI search work in the user's interest and not in the interest of the search engine provider? Can we break free from being the product and become the user in the mix…
For the subscribers or first-time readers (welcome!), thanks for joining! A short general intro: I am Iskander Smit, educated as an industrial design engineer, and have worked in digital technology all my life, with a particular interest in digital-physical interactions and a focus on human-tech intelligence co-performance. I like to (critically) explore the near future in the context of cities of things. And organising ThingsCon. I call Target_is_New my practice for making sense of unpredictable futures in human-AI partnerships. That is the lens I use to capture interesting news and share a paper every week.
Notions from the news
I am trying to tweak the format a bit. The notions from the news are described in a running text below, and I have selected some overarching articles in the introduction. Curious to hear what you think of course!
Amy Webb is a respected futurist who produces extensive trend reports, usually in March during SXSW. She sees a technology supercycle happening with AI.
Ethan Mollick
Ethan Mollick is making a good point that the impact of current and near-future AI developments is already expected to be significant; we don’t need superintelligence for this impact.
Tim Brinkhof
Simon de la Rouviere is linking autonomous things and new cryptography. LLM AI got involved with memecoins, which are living independently on Twitter.
Simon de la Rouviere
Venkatesh Rao closed his Ribbonfarm blog and continues with Contraptions. In this post he explores that concept through thinking about machines that slip on banana peels. Metaphorically.
Venkatesh Rao
Human-AI partnerships
Meta is well-positioned to benefit generative AI; that is the TL;DR of a long-read analysis by Ben Thompson.
The balance of human and AI working together also has another aspect next to the quality and value of the output; there is also an element of trust and accountability. Cory Doctorow shows the aspects and (legal) consequences of “the human in the loop”.
The researcher sees good results in using AI as a qualitative researcher.
Or you build a military system…
What if you delegate all decisions for a week to AI? Kashmir Hill took a decision holiday.
Robotic performances
I tend to avoid the humanoids as the interesting robot category, but there is also an interesting question: what makes a humanoid a humanoid? Is it more about being a human as much as possible? Or are these Boston Dynamics robotic factory employees just optimized with human characteristics, just another iteration of factory robots, not a new humanoid category?
And can robotic things (like cars) become more valuable by the conversational AI?
Does a wheelchair need wheels?
You would expect that AI recognition would always be part of driverless cars.
The narrative can be crucial in shaping futures insights. The impact of autonomous vehicles is not new; focusing on the windowless car is nice. Or will goggles make more sense after all?
Immersive connectedness
Is it just a new updated device, that new tiny Mac Mini, or a sign that more powerful machines are also needed to support the move to intelligent houses, where the laptops are for on the road? Is home computing a new iteration of home intelligence?
Sometimes (or maybe even often) are, the first movers, not the best equipped for changing realities. The Amazon Echo and similar devices are not built for hallucinations; the expectations differ. Will it indeed be doomed for the graveyard, or is a resurrection still possible?
The returning disappointment that is called USB-C.
Tech societies
In case you missed it, visions of the future for AI in society are a thankful topic to profile yourself for the leading generative AI companies. It makes sense, and it might also be a sign of the clear shift happening that these tools are more and more embraced by the existing big tech, and the newbies need to keep it visible. Next up is a piece on regulation.
For how long do we say “data is the new oil”. In value, that is happening for the first time now.
How will conservative presidents deal with this new world? You know the answer.
Paper for the week
The paper mentioned in the triggered thoughts is "When combinations of humans and AI are useful: A systematic review and meta-analysis”
Inspired by the increasing use of artificial intelligence (AI) to augment humans, researchers have studied human–AI systems involving different tasks, systems and populations. Despite such a large body of work, we lack a broad conceptual understanding of when combinations of humans and AI are better than either alone.
Vaccaro, M., Almaatouq, A. & Malone, T. When combinations of humans and AI are useful: A systematic review and meta-analysis. Nat Hum Behav (2024). https://doi.org/10.1038/s41562-024-02024-1
Looking forward
The focus will be working on projects and preparing ThingsCon, as mentioned above. If you are looking for some events, ICAI is looking into enhancing human interaction with AI, which is triggering some questions on what the right angle… There is a conference on Enterprise UX that some old friends are organizing and hosting. Probably a good experience if the topic is your thing. A different cookie is the yearly Glow light festival starting this weekend. The smart city crowd is in Barcelona. Internet culture is explored in different aspects at v2 during Crash Course.
Enjoy your week!
