but google did erode trust in their product, and the american government went after them, so they "made it more obvious" which still really didnt change that most people dont care if something is sponsored and just look at the first result.
it's something that continually needs to be reenforced again and again. somebody will be made example of.
you dont know what you dont know. walking into hockey for the first time, you may think you should be looking for the puck.
but really, what you want to look for is how the players are moving. it's sort of a "which one is different from all the others." one person will clearly be moving in a completely unique way, as the others chase them or vie to get open or get in somebodys way. to acomplish this identification, youre looking at their legs, shoulders, hands, feet, and heads.
The same applies to context vs a database. If a reasoning model makes a decision about something, it should be put off to the side and stored as a value/variable/entry somewhere. Instead of using pages and pages of context, it makes sense for some tasks to "press" decisions that become more permanent to the conversation. You can somewhat accomplish that with notebooklm, by turning results into notes into sources, but notebooklm is insular and doesnt have the research and imaging features of gemini.
And also, in writing, writing from top to bottom has its disadvantages. It makes sense to emulate human writing process and have passes, as you flesh out, and conversely summarize writing.
Current LLMs can brute force these things through emulation/observation/mimicry but they arent as good as doing it the right way. Not only would I like to see "skills" but also "processes" where you create a well defined order that tasks are accomplished in sequence. Repeatable templates. This would essentially include variables in the templates, set for replacement.
> Not only would I like to see "skills" but also "processes" where you create a well defined order that tasks are accomplished in sequence. Repeatable templates. This would essentially include variables in the templates, set for replacement.
You can do this with Gemini commands and extensions.
The template would more define the output, and I imagine it more recursively.
Say we are building a piece of journalism. First pass, do these things, second pass build more coherent topic sentences, third pass build an introduction.
Right now, the way that models write from top to bottom, the introduction paragraph seems to inform the body, and then the body is just a stretched out version of the intro. Whereas how it should work is the body is written and then condensed into topic sentences and introductions.
I find myself having to baby models, "we are going to do this, lets do the first one. ok now lets do the second one, ok now the third one. you forgot the instructions, lets revise with the parameters you were given initially. now lets put it all together."
I'm babbling, I just think these interfaces need a better way to define "lets write paragraph 4 first, followed by blah blah" to better structure the order in which they tackle tasks.
it's not unrealistic to be selecting for people with strong language skills and the ability to break tasks into discrete components and assemble them into a process. or the skill of being able to define what they do not know.
a lot of what makes a person good with an llm makes them also good at general problem solving.
not just the keyboard either, but the text editor box (or address bar /search) in general. i cant count the number of times i try and put the cursor before a word, i see it is before the word, i let go, and the cursor moves to the end of the word. if i wanted it at the end of the word i would have put it there before letting go.
also, the damn period next to n in the address bar. no i didnt mean to type every word in a sentence with a period delimiting between words.
It has the exact same bug as mentioned above. I solely use the spacebar for cursor movement, and the cursor returns to the end of the line/word at random times. I couldn’t find a pattern when it happens. It’s especially annoying when it happens with something long like a long path in a URL bar.
> i see it is before the word, i let go, and the cursor moves to the end of the word. if i wanted it at the end of the word i would have put it there before letting go.
Having never implemented something like this, I wonder if the algorithm could take into account how long the cursor lingered on each position before being let go. If it spent significantly longer in a position before the word, and your finger happens to move a little bit when you let go, that slight movement shouldn't affect the cursor position.
Apple is usually pretty good about this stuff but they've really been slipping on the keyboard.
Facebook's stock was up 20% later in the year after the acquisition.
Facebook was worth $134.2-139.2B end of 2013 and $217.5-218.5B end of 2014.
I would say it is misleading to say it cost them $15B in shares if the remaining shares FB kept ended up more valuable after the transaction.
reply