Saved

Much of these problems with Google today stem from a lack of visionary leadership from Sundar Pichai, and his clear lack of interest in maintaining the cultural norms of early Google. A symptom of this is the spreading contingent of inept middle management. Take Jeanine Banks, for example, who manages the department that somewhat arbitrarily contains (among other things) Flutter, Dart, Go, and Firebase. Her department nominally has a strategy, but I couldn't leak it if I wanted to; I literally could never figure out what any part of it meant, even after years of hearing her describe it. Her understanding of what her teams are doing is minimal at best; she frequently makes requests that are completely incoherent and inapplicable. She treats engineers as commodities in a way that is dehumanising, reassigning people against their will in ways that have no relationship to their skill set. She is completely unable to receive constructive feedback (as in, she literally doesn't even acknowledge it). I hear other teams (who have leaders more politically savvy than I) have learned how to "handle" her to keep her off their backs, feeding her just the right information at the right time. Having seen Google at its best, I find this new reality depressing.

In the first approach, the primary interface affordance is an input that directly (for the most part) instructs an AI model(s). In this paradigm, people are authoring prompts that result in text, image, video, etc. generation. These prompts can be sequential, iterative, or un-related. Marquee examples are OpenAI's ChatGPT interface or Midjourney's use of Discord as an input mechanism. Since there are few, if any, UI affordances to guide people these systems need to respond to a very wide range of instructions. Otherwise people get frustrated with their primarily hidden (to the user) limitations.

The second approach doesn't include any UI elements for directly controlling the output of AI models. In other words, there's no input fields for prompt construction. Instead instructions for AI models are created behind the scenes as people go about using application-specific UI elements. People using these systems could be completely unaware an AI model is responsible for the output they see.

The third approach is application specific UI with AI assistance. Here people can construct prompts through a combination of application-specific UI and direct model instructions. These could be additional controls that generate portions of those instructions in the background. Or the ability to directly guide prompt construction through the inclusion or exclusion of content within the application. Examples of this pattern are Microsoft's Copilot suite of products for GitHub, Office, and Windows.

they could be overlays, modals, inline menus and more. What they have in common, however, is that they supplement application specific UIs instead of completely replacing them.

complexity can make it difficult for both domain novices and experts alike to learn how to use the interface. LLMs can help reduce this barrier by being leveraged to prove assistance to the user if they’re trying to accomplish something, but don’t exactly know how to navigate the interface.The user could tell the program what they’re trying to do via a text or voice interface, or perhaps, the program may be able to infer the user’s intent or goals based on what actions they’ve taken so far.Modern GUI apps are slowly starting to add in more features for assisting users with navigating the space of available commands and actions via command palettes; popularised in software iA Writer and Superhuman.

for executing a sequence of tasks as part of a complex workflow, LLM powered interfaces afford a richer opportunity for learning and using complex software.The program could walk them through the task they’re trying to accomplish by highlighting and selecting the interface elements in the correct order to accomplish the task, along with explanations provided.

Expert interfaces that take advantage of LLMs may end up looking like they currently do - again, complex tasks require complex interfaces. However, it may be easier and faster for users to learn how to use these interfaces thanks to built-in LLM-powered assistants. This will help them to get into flow faster, improving their productivity and feeling of satisfaction when using this complex software.

unlike Clippy, these new types of assistant would be able to act on the interface directly. These actions will be made in accordance to the goals of the person using them, but each discrete action taken by the assistant on the interface will not be done according to explicit human actions - the goals are directed by he human user, but the steps to achieve those goals are unknown to the user, which is why they’re engaging with the assistant in the first place

If I had to summarize it, I would say that the signal to noise ratio is what wore me down. We start companies to build products that serve people, not to sit in meetings with lawyers.  You need to be able to answer the "what have I done for our users today" question with "not much but I got promoted" and be happy with that answer to be successful in Corp-Tech.

being part of a Corporation means that the signal to noise ratio changes dramatically.  The amount of time and effort spent on Legal, Policy, Privacy - on features that have not shipped to users yet, meant a significant waste of resources and focus. After the acquisition, we have an extremely long project that consumed many of our best engineers to align our data retention policies and tools to Google. I am not saying this is not important BUT this had zero value to our users. An ever increasing percent of our time went to non user value creation tasks and that changes the DNA of the company quickly, from customer focused to corporate guidelines focused.

the salaries are so high and the options so valuable that it creates many misalignments.  The impact of an individual product on the Corp-Tech stock is minimal so equity is basically free money.  Regardless of your performance (individually) or your product performance, you equity grows significantly so nothing you do has real economic impact on your family. The only control you have to increase your economic returns are whether you get promoted, since that drives your equity and salary payments.  This breaks the traditional tech model of risk reward.

The Ai Pin makes the same conceptual mistake behind all the assistants that preceded it: to treat all people as if they were so utterly helpless and clueless to manage even basic stuff. And to grossly miscalculate which tasks people find tedious and willing to delegate to a machine. These assistants want to assist with stuff people have no problem doing themselves, and they do so through an interaction model that ultimately makes things more awkward, impractical, and longer to accomplish. (On the other hand, it’s a good interaction model for people who have different types of motoric or visual disabilities and need assistance when sending and receiving messages, collecting information, etc.).

Instead there’s this urge to create The Next Big Thing that will be a hit for everyone, everywhere. And to create it in one fell swoop, skipping all the steps that might help you really get there.

Despite its direct affiliation, Stripe Press provokes a distinctive, emotional feeling. It’s an example of how form affects soft power. By focusing on actual, physical books — and giving them a loving, literary treatment — Stripe shows this project is firmly outside the world of “marketing.” Rather, this is a place for Stripe to demonstrate its ideological affinities and reinforce its philosophical positioning. The affection this project has earned suggests it has found distribution.

Most obviously, they can invest in it via in-house initiatives. Even moderately sized tech companies have large marketing teams capable of running interesting experiments, especially if augmented with external talent. Business banking platform Mercury has made strides in this area over the past couple of years, launching a glossy, thoughtful publication named Meridian.

Marvel licensed the filming rights of the X-Men and related concepts, such as mutants, to 20th Century Fox, who created a film series based on the franchise. Years later, Marvel started their own film franchise, known as the Marvel Cinematic Universe (MCU), which focused on characters that they had not licensed to other studios (see below). At the time, the rights to Quicksilver and Scarlet Witch were disputed by both studios. As they both held the rights to the characters, with Fox citing the characters' mutant status and being children of Magneto and Marvel citing the twins' editorial history being more closely tied to the Avengers rather than the X-Men, the studios made an agreement wherein both of them could use the characters on the condition that the plots did not make reference to the other studio's properties (i.e. the Fox films could not mention the twins as members of the Avengers while the MCU could not mention them as mutants or children of Magneto).[215] The arrangement became moot following the acquisition of 21st Century Fox by Disney – the parent company of Marvel Studios, and the confirmation that future X-Men films will take place within the MCU.

From the start, Beyoncé preaches her desire to create a “safe space.” “Renaissance means a new beginning,” she says; it’s a balm “after all we’ve been through in the world.” But what exactly is she referring to? The onslaught of death and illness brought on by the continuing pandemic? The laws aimed at criminalizing trans children and adults? The rising misogyny, homophobia, and anti-Blackness that leads to grave violence? The various, ongoing genocides? Beyoncé gives us no context for what she’s referring to or how it touches the shores of a life dominated and driven by the kind of wealth that insulates her from harm. Her words reflect broadly liberal pablum meant to give the appearance of care and mean just enough that her fans can project radicalness upon her but not so much that she would ruffle anyone enough for her to lose money or be forced to stand for something.

there is no star of such magnitude who more cunningly positions themselves as apolitical than Beyoncé. Her performance as an icon is meant to connect with the broadest number of people possible. To do that, her refusal to stand for anything specific beyond the watered-down treatises on Black excellence must be maintained.

More than anything, Renaissance is a testament that Beyoncé is a brand that stands for absolutely nothing beyond its own greatness

In the near future, transforming text will become as commonplace as filtering images. A new set of tools is emerging, like Photoshop for text. Up until now, text editors have been focused on input. The next evolution of text editors will make it easy to alter, summarize and lengthen text. You’ll be able to do this for entire documents, not just individual sentences or paragraphs. The filters will be instantaneous and as good as if you wrote the text yourself. You will also be able to do this with local files, on your device, without relying on remote servers.

Initially, many of Photoshop’s capabilities were adaptations of analog effects. For example, “dodge” and “burn” are old darkroom techniques used to alter photographs. There are countless skeuomorphic names throughout digital image editing tools that refer to analog processes.

Text seems like it would be easier to manipulate than images. But languages have far more rules than images do. A reader expects writing to follow proper spelling and grammar, a consistent tone, and a logical sequence of sentences. Until now, solving this problem required building complex rule-based algorithms. Now we can solve this problem with AI models that can teach themselves to create readable text in any language.

With iA Writer 7 you can manually mark ChatGPT’s contributions as AI text. AI text is greyed out. This allows you to separate and control what you borrow and what you type. By splitting what you type and what you pasted, you can make sure that you speak your mind with your voice, rhythm and tone.

As a dialog partner AI makes you think more and write better. As ghost writer it takes over and you lose your voice. Yet, sometimes it helps to paste its replies and notes. And if you want to use that information, you rewrite it to make it our own. So far, in traditional apps we are not able to easily see what we wrote and what we pasted from AI. iA Writer lets you discern your words from what you borrowed as you write on top of it. As you type over the AI generated text you can see it becoming your own. We found that in most cases, and with the exception of some generic pronouns and common verbs like “to have” and “to be”, most texts profit from a full rewrite.

we believe that using AI for writing will likely become as common as using dishwashers, spellcheckers, and pocket calculators. The question is: How will it be used? Like spell checkers, dishwashers, chess computers and pocket calculators, writing with AI will be tied to varying rules in different settings.

We suggest using AI’s ability to replace thinking not for ourselves but for writing in dialogue. Don’t use it as a ghost writer. Because why should anyone bother to read what you didn’t write? Use it as a writing companion. It comes with a ChatUI, so ask it questions and let it ask you questions about what you write. Use it to think better, don’t become a vegetable.

Once you come to the realization/acceptance that no one – not even the interested, motivated, and committed — has the kind of focused attention available for your thing that you assume they have, you will have a much better chance of capturing and sustaining any of it at all.

Ask as few actions of your audience as possible, and provide as little information to support those actions as possible. Everything else that you could put on your screen is a distraction.

This paper maps concepts from AI alignment onto a basic, three step interaction cycle, yielding a corresponding set of alignment objectives: 1) specification alignment: ensuring the user can efficiently and reliably communicate objectives to the AI, 2) process alignment: providing the ability to verify and optionally control the AI's execution process, and 3) evaluation support: ensuring the user can verify and understand the AI's output.

the notion of a Process Gulf, which highlights how differences between human and AI processes can lead to challenges in AI control.