Tag: ai

After launching its Comet AI web browser exclusively for paying subscribers, Perplexity has now made it downloadable for everyone. You can download the browser from the official website for testing, even if you do not have a Perplexity account.

However, Perplexity limits functionality of the browser in that case. First, Comet is a Chromium-based browser. That is good on the one hand, as it offers similar performance and compatibility as Google Chrome. It also means that you can install browser extensions that are compatible with Chrome in the desktop version.

You can import your last browsing session from most other browsers on first start, but can skip it as well. Afterwards, Perplexity displays a few options to you, which most users may want to disable. This includes autostarting the browser with the operating system, making it the default browser, or sending telemetry data to Perplexity.

The browser opens and it looks like Chrome for the most part. That is to be expected, as all Chromium-based browsers look pretty much the same. The new tab page displays a few widgets, like the time or Nvidia’s stock price. The star of the show is the text prompt at the top. You may use it to interact with Perplexity’s AI.

Comet Assistant displays a few suggestions when you activate the text field, such as “Find emails from the last 7 days that require replies and draft responses”, or “find top rated noise-cancelling headphones under $300 and open product pages from three major retailers with the lowest prices”.

So, it is basically an AI chat, but this AI has agentic capabilities. It is rather limited for free users, as pro searches and several other features are not available in that case.

The browser displays its results on Perplexity’s website. In fact, you may get similar results when you use the company’s website directly to interact with the AI.

You may also notice that you get constant requests to sign in, if you do not. Open a new tab page, get a request. Similarly, prompts may be displayed on the results page.

If you do not sign-in, you get access to base AI features only. You can’t, for example, change the AI model that the browser uses to process queries. There is also no option to upload files or photos for processing, or run deeper queries.

However, results looked fine in quick tests, even non-Pro searches. Whether the information is accurate is another topic. As is the case with all AI, you should verify important information before you act on it.

Perplexity is Chromium-based. Which is good. You can open chrome://flags to tinker with advanced preferences, and get most of the settings of the browser as well.

Now the big question that I still have regarding all these AI browsers is about the “why” and whether that why warrants paying a monthly subscription fee. Why should I use such a browser? I do not really see the killer-application for it right now. Yes, it can run deep research tasks for you or perform some tasks for you, but how useful is that really, especially if you prefer to stay in control the whole time?

Anyway, you can give Perplexity a try, if you want.

When the Browser Company launched its Arc web browser, it was heralded by part of the media as the evolution that browsers needed. Then, the Browser Company announced that it would discontinue its browser to focus on another. This new browser, called Dia, would be an AI-browser first and foremost.

It was not really clear why, but the recent announcement of a Pro subscription plan could shed some light into the plans.

Dia, which is available in early testing for certain Mac devices only, is free to use. The AI parts have limits, however, and the newly announced Pro subscription unlocks unlimited access to the AI.

For $20 per month, users gain unlimited access. The first commercial subscription plan is just the beginning, as The Browser company plans to introduce additional plans that may cost between $5 and several hundred Dollars per month.

The price of a Pro subscription is just shy of the price of a ChatGPT Plus or Google AI Pro subscription. Granted, Dia appears to offer tighter integration in a browser.

Dia offers integration of a chatbot, but one feature that sets it apart from competing browsers currently is skills. These are shortcuts for repetitive tasks. It also supports cross-tab analysis and content synthesis. Is that enough to warrant a subscription? Only time will tell, but my initial take is that it will be very difficult for the company to turn a profit, unless it bakes something into the browser that is truly useful.

Now You: would you pay for a browser? What is your take on the $20 per month price of a subscription? Feel free to leave a comment down below.

Mainstream AI tools rely on large sets of data that they are trained on. This data, often websites, posts, comments and the like, has been created by humans (for the most part).

Reports suggest that part of the Web population is shifting towards using AI first for any of their tasks and there are good reasons (and some not so good ones) for that.

On the positive side, it is dead simple. You get an answer right away and do not have to wade through different websites in the hope that they offer the answer. Links are provided to verify the claims. And this works reasonably well for many requests.

The downside to this is that AI is still hallucinating. In plain terms, AI may return information that is inaccurate, plain wrong, or created out of thin air. That is bad, especially if the information could end up in disaster. My bet is that most users are unaware of this and take information at face value. It is also changing how the Web is used.

The consequences are severe for website publishers who rely on ads, at least those who do not have the pull to sell their data to AI companies. Traffic will drop. Google is pushing AI overviews, and this leads to less visits. Google says that this is not the case, that searchers click on links in AI overviews, but this appears exaggerated.

In the next couple of years, websites will receive less and less traffic from search. All search engines move to AI content and the reason is simple: this keeps users on their site. Means, they get to keep all the ad Dollar.

Google just introduced its AI into Chrome Canary. This too will keep users inside Chrome and Google’s vast network of services.

Granted, some sites will not be affected by this. If you offer a service that AI can’t replicate, you are likely golden. Think of downloads, streaming media, good journalism, user interactions, games and more.

The traditional newspaper or blog will have a hard time, unless it is very special. Other services, too. Think of all those flight and hotel booking sites. Agentic AI might do that directly in the browser or on a website, no need to go through dozens of sites yourself.

Yes, aggregators like Booking or Trip exist already, but AI can replicate that without issues.

Many will stop publishing or be pulled from the Internet entirely. Less traffic means less ad Dollar means less money for paying writers, hosting and administration. It is a downward spiral.

AI won’t know that certain “things” exist

But this also means less content that AI can vacuum up. If no one writes about a certain topic anymore, AI won’t know about it. It will be as if it does not exist at all.

So, when the last independent games review site shuts down, AI may not have any info about the latest games anymore. Yes, this is a bit exaggerated, as there may be reviews on Reddit or other community sites.

Still, less content means less nuanced information about a topic.

Closing Words

If you operate a website, unless you are a big corpo with AI deals already, then you might want to search for different monetization options. Reliance on ads may not cut it anymore in the future. Expect more sites to go subscription-based. Some might ask for donations. Fatigue will kick in and money can only be spend once on something.

Again, I may be a bit too pessimistic about the outlook, but the Web as we know it is changing dramatically.

Now You: what is your take on AI taking over the world and destroying the Internet as we know it? Let me know in the comment section below.

How do many of the AI startups make money? Many have launched products that are freely available and without ads. There are multiple answers to the question, depending on the business’ strategy. Some want growth at all costs to sell the business later on. Others plan to introduce ads or subscriptions at a later point in time.

Soon, users from all over the world will see more ads in chats with AIs, as Google is expanding its Adsense program to chats.

Bloomberg reports that Google is expanding Adsense to keep its edge in the advertising vertical. Tests with select startups such as iAsk and Liner have begun already reportedly.

New types of AI-powered services spring up virtually everywhere on the Internet. Liner, for example, is an AI-powered search engine, while iAsk promises answers to questions that users may have.

Most services have in common that they are not page-based websites. Blogs or regular websites publish static pages most of the time, whereas interactions with AI are usually dynamic.

While companies may display Adsense on the pages, they were not particularly optimized for the queries of users. This changes with the rollout of the new Adsense format.

Integration means that ads may show up in chats. Where and how depends on the AI service, but users should expect to see an increase in advertisement, especially on AI sites that are not operated by billion Dollar companies.

Good news is that content blockers should be able to deal with those ads, just like they are regarding ads in search or on websites.

Now You: do you use AI services regularly? If so, for what purpose and what do you like or dislike in particular when you compare the AI service to other means of acquiring the information?

What is the proper etiquette when chatting with AIs? Do you say “thank you”, “please” or “see you next time”, or do you prefer the direct approach and avoid any form of politeness when communicating with AI tools?

OpenAI CEO Sam Altman was recently asked about the electricity costs that these pleasantries would generate. According to him, processing these cost “tens of millions of dollars”.

If everyone would stop being nice to AI, at least in this regard, AI companies electricity bills would go down noticeably. While that has no immediate benefit to the user, it is at the very least beneficial to the environment.

Altman ended the answer with a joke, stating that the money was well spend as “you never know”, referring to the possibility of an AI uprising and the potential saving of people who have been nice to AI in the past.

You might say that AI causing havoc to bring doom to the human race might not be persuaded to let humans live based on their use of “please” or “thank you”, and you are probably correct here.

Where does this leave the statement? When you avoid please or thank you, you will help AI companies save power, which is beneficial to them, but also the environment, at least potentially.

So, AI systems seem to be almost everywhere already or are in the process of being integrated into pretty much any device or product.

OpenAI, maker of ChatGPT, has now announced that you may call ChatGPT to have a chat with the AI.

Call 1-800-ChatGPT, or 1-800-242-8478, to do so. You should note that you can only do so with a US or Canada number. If you live outside those two countries, you may be able to message the number via WhatsApp to at least have a chat on Meta’s messenger instead.

OpenAI has limited both options. Calls are limited to 15 minutes per month, presumably per phone number, and there is also a daily limit on WhatsApp, but that is not mentioned in the announcement.

Another restriction is that the cutoff-date is October 2023. This means that this particular version of ChatGPT does not know anything about events that happened after October 2023.

Privacy-conscious users should be aware that OpenAI stores conversations and messages, and that it may also review them. OpenAI says this is for a limited time only and only for “safety and abuse prevention purposes”. It is also only available for ages 13 and up, but it is unclear how OpenAI wants to make sure that the age limit is enforced.

Fax or Letters next?

With telephone conquered, Fax would be the next logical step for OpenAI to offer its services on. It might not be the most elegant solution, considering that it might take a while to get replies, but it would make for a great headline and do wonders for publicity.

Letters might also be an option. While it would take some serious time to get an answer or do a back and forth with the AI, it could be a great option for digital-detoxing.

Have you chatted with ChatGPT or another AI in recent time? What is your take on these systems? Useful addition or more off a plaything will little value? Feel free to leave a comment down below.