Amazon lets advertisers use generative AI to make product shots more lifelike

For advertisers without a graphic design or creative team, this is a quick solution to spruce up images that would otherwise be a bore

Complement this with auto-generated product descriptions, and Amazon might be one of the only Big Tech company making the most impactful “under the hood” AI improvements for their customers. That is, they are not making fancy bots or automatic PowerPoints and over-doing the “we’re riding the AI wave” thing.

Also, to note:

Amazon says putting products in a lifestyle scene can lead to 40 percent higher click-through rates 

The Verge

A Product Model to help PMs communicate clearly

I dream of a world where a Product Model can be generated from the organization of the backend files that respect the Domain-Driven Design (DDD) approach. That would quickly enable Product people to formalize a shared language from real code, and not spend hours philosophizing about why X should be named Y. Alas, this time has yet to come.

Indeed, one of the hard things to do in product development is to come up with a naming convention for your app that makes sense to everyone. Engineers, designers, and everyone else on the team need this shared language to talk about the product.

Another thing that is important for Product Managers is the ability to quickly conceptualize requirements for a feature or a product. Sometimes, user stories are too vague or incomplete.

Taking free inspiration from DDD, which states that in naming, the engineering logic must be as close as possible to the real-world, business logic, we can imagine the following model:

Domain Layer

This is where we name the entities and concepts that power the product. Domains are where the unique business logic is expressed. Say you are building an ecommerce app, you might have Cart, Accounts or Payments. If you’re building an AI companion dating app, you might have Profile or Chat.

Infrastructure Layer

To me, this is where the backend orchestrates services that talk back and forth with the database. This is where I might refer to purely technical, non-visual elements such as a matching algorithm or a dedicated LLM logic.

Application Layer

This is the traditionally used layer when Product people think about products. This can capture interaction-based, user-centered flows that blend different objects from the Domain and Infrastructure layers, such as Sign Up or Add Item to Cart. 

In short, because software applications are complex systems, thinking about the Application layer is not enough.

Product people should take into account the impact of a new feature in conceptual terms (Domains) and in technical, non-visual terms (Infrastructure).

Some Tips to Detect Fake News

Two things to keep in mind:

Remember the business model of media entities. Beneath the noble intention of sharing information, the business model of media companies is to maximize the number of people seeing ads on their properties, so they maximize traffic. How? Boring stuff doesn’t drive traffic, but drama, and thus exaggeration, does.

Remember the goal of arguments, even online. A successful argument refutes the central thesis of the other party’s argument. Beware of fallacies and biases like personal attacks (ad hominem), misrepresenting/exaggerating the opponent’s argument to make it easier to attack (straw man fallacy), or saying that something is true because it’s supported by an authority (appeal to authority). There are many more.

Without further ado, here are the tips: 

  1. Check the source. A random Twitter/X account with 13 followers can share a post with 1M views. But it doesn’t mean that the content is legit. Even accounts with a lot of followers are unreliable. One example is The Spectator Index (2.5M followers) which always exaggerates its claims for more views. Check the replies, check the origin of the information. Who shared it first? Yes, reputable sources are very hard to find. A heuristic I use is that if the content is sensationalistic and without context, there is a good chance that the “fact” is being exploited.
  2. Cross-check with other sources. If only one source shares the information, and you feel that they might not be a primary source, then exercise caution. One source is not enough to guarantee the validity of the information. Consider checking other sources. 
  3. Check dates & times. A lot of people discard checking the date or time of content they share. Don’t be like them! Check the date & time. 
  4. Beware of emotional manipulation. Maybe I’m weird but my radar is triggered when the content shared is particularly dramatic (physical abuse, suffering children). Indeed, whenever we see suffering, our rational mind shuts off, and we are outraged (this is good) so we externalize it and forget to check the source, cross-verify, or check the date (this is less good). 
  5. Ask the Internet for verification. There’s a useful Discord server. Check: They are an Open Source Intelligence community. You can post a piece of info on the Discord and have people try and verify it. 

Remember to take a break from reading all of these dreadful news. Surely humans are not meant to process such a large amount of vivid violent imagery. 

The CIA Read Allies Encrypted Communications — For Decades

In a surprising, ahem, move, the CIA owned the one company that numerous countries trusted for their cryptography needs: 

For more than half a century, governments all over the world trusted a single company to keep the communications of their spies, soldiers and diplomats secret.

The company, Crypto AG, got its first break with a contract to build code-making machines for U.S. troops during World War II. Flush with cash, it became a dominant maker of encryption devices for decades, navigating waves of technology from mechanical gears to electronic circuits and, finally, silicon chips and software.

But what none of its customers ever knew was that Crypto AG was secretly owned by the CIA in a highly classified partnership with West German intelligence. These spy agencies rigged the company’s devices so they could easily break the codes that countries used to send encrypted messages.

Read more here.

Anne L’Huilier, Pierre Agostini, and Feren Krausz won the 2023 Physics Nobel Prize. They developed “attoseconds”, extremely short pulses of light where one attosecond is one billionth of a billionth of a second. 

This allowed them to study the incredibly fast movements of electrons within atoms and molecules in real time. Before, it was impossible to observe such rapid movements because of their fleeting nature.

We now have a way to investigate the fundamental behaviors of electrons, which can enable a variety of technological advancements such as faster electronic devices and inroads in chemistry and biology. 

Two further fun facts: Anne L’Huilier started working on this in the 1980s and got back to the amphitheater to continue her class after she got a call informing her she had won. Indeed, her greatest gift to humanity is her teaching.

Why Apple Will Not Release a Google-like Search Engine

It is reasoned that comments from Apple SVP of Services Eddy Cue saying Google’s search is the best and that Apple has no incentive to make its own are probably true, but could also be a measure to try and protect Google from government enforcement.

A rumor has it that Apple is developing its own search engine to compete with Google. This is true, as Apple must create some crawling/ranking software for Siri, Spotlight and such. However, I do not believe they are going to release a proper search engine. The reason for this rumor could be related to the anti-trust lawsuit against Google, and Apple is defending them. The incentives for the deal are too good for Apple, they are getting $15B a year for almost 0 effort. 

On Voice in AI Companions

Those of us who are blessed to have many close friends and family members in our life may look down on tools like this, experiencing what they offer as a cloying simulacrum of the human experience. But I imagine it might feel different for those who are lonely, isolated, or on the margins. On an early episode of Hard Fork, a trans teenager sent in a voice memo to tell us about using ChatGPT to get daily affirmations about identity issues. The power of giving what were then text messages a warm and kindly voice, I think, should not be underestimated.

Good insight from Casey Newton.