I think it would do the discourse some good if we stop thinking about the value of genai as binary (useful/useless). Clearly some people find it useful, others don't (for now).
I'm more interested in GenAI's "Google" moment. What will be the innovation that adds value for *all* users?
When the tools fail the ordinary consumer, they fail (period).
Where the power users (and developers) go wrong is in flippantly dismissing the legitimate concerns of the ordinary users.
We aren't talking about specialized professional tools like Photoshop either. General purpose tools like ChatGPT are marketed to the ordinary consumer.
But for everyone else, it sucks. The value of genai can't be realized or appreciated when the tools are broadly considered unusable, time consuming, and not trustworthy.
And again, like with early search tools, generative AI can be amazing when you know how to use it. It is objectively more powerful and more efficient than pre-genAI info gathering tools.
Power users recognize this and hence are more willing to engage with the steep learning curve to get there.
What they don't tell you is that you actually have to do a lot of cognitive work behind the scenes to make the wizard useful.
And so we see the divide again. For power users, this is no big deal.
And gen AI users are right to point that out.
Like the early search engines, model developers market their tools as all knowing "intelligent" wizards that hold the answers to anything and everything.
Plus, you have to have some basic knowledge of the subject you're asking about because now you also have to be able to catch when the machine is confidently wrong or making things up.
This is non-trivial work compared to traditional research.
Like with the early search engines, it takes work to get good results from genAI.
You not only have to write prompts using the formalities and syntax that the model expects, but you also have to know what a good result (to you) looks like and you have to explain that in your prompt.
I think we're at a similar juncture with Generative AI. It's amazing for power users and not so amazing for everyone else, and that's why there's this tension in the discourse over its value.
(and Google would eventually win the search engine wars)
Google was the first search engine to emerge that recognized this tension and fixed it. Instead of forcing users to adjust to the technology, Google adjusted to the users.
Ordinary users could now just ask their questions like ordinary people sans the formalities AND get good results.
Power users said "skill issue." Ordinary users were unamused.
They felt they shouldn't have to learn a new language and take on most of the cognitive load for technology that promised to make their lives easier and more efficient.
They were not necessarily wrong to think that either.
Plus, it certainly didn't help that many of these early search engines marketed themselves to ordinary consumers as seamless Internet research machines (one literally featured a butler!)
Ordinary users weren't getting what they expected (or were promised).
en.wikipedia.org/wiki/Ask.com...
But not all search engine users were power users.
For many, the promises of a powerful search wizard that could find the answer to any question you asked fell flat. Users were typically met with irrelevant results (and oftentimes nothing at all).
Now for power users, this was no big deal. They were "query literate" and they accepted the labor required to yield good results.
Because when used effectively, search engines were incredible and immensely efficient compared to any offline information gathering analog.
Instead you might try: "dehydration symptoms"
Skim the results
Adjust the query: "dehydration symptoms OR signs"
Repeat
(you also had to be willing to try different search engines too!)
For example, you couldn't ask an early search engine: "what are the symptoms of dehydration?"
For the best results, searches needed to be stripped to a few core terms, use Boolean operators, and you had to be willing to try multiple variations of your query.
Most queries yield bad results.
Early indexing and search tools were a game changer, but you had to know how to use them (again, think pre-Google).
Query literacy was the dividing line between power and ordinary users. You had to know how to ask your question to get good results (a tremendously steep learning curve for most).
Meanwhile, Internet developers (or "hucksters" as this article called them) touted its value.
Books, newspapers, classrooms, business meetings, etc would all one day be online. Our offline inconveniences would be a thing of the past: share.google/oVBgGvn9L7Ve... share.google/affNa75cEXQt...
The pre-Google Internet was considered by many to be unusable, and similarly, a time suck.
Finding answers to trivial questions meant sifting through a web of unstructured data and random amateur pages full of misinfo and low value content. Early users often complained about info overload.
On one side of the gen AI debate are the skeptics who say it adds unnecessary friction to trivial tasks that creates more work for the user and saves no meaningful time.
On the other side are the power users who suggest the skeptics need to learn how to use the tools.
I think both are right. 🧵
The idea being that suicide is a conscious and independent and usually competent decision of an individual that supersedes causation.
The history of liability (re: your latter point) is very interesting (and complicated). In tort, courts are mostly hesitant to assign liability to another for "causing" suicide, even in some of the most extreme cases of intentional infliction of emotional distress.
YUPPP the lemonade stand girl is exactly what runs through my head every time I write on this topic. IT SHOULDN'T NEED TO BE THIS WAY.
Yes! So, I have a much longer article where I discuss this. Suicide is viewed through the lens of societal and environmental issues that policymakers have long ignored. We have to fix those problems. But in the meantime, people need something. To the extent AI is harm reduction, it's something.
This is the right answer.
Me holding my 230 coin. The coin says protecting the web, 230, CDA Section 230 x30 since 1996.
Selfie of me holding the section 230 coin
My @techdirt.com Section 230 coin!! Just in time for Coachella too 🌵🙂↔️🌴
