Better Browsing to avoid issues GPT4 with Browsing via Bing

So I’ve done some tests with GPT4 with Browsing via Bing today,

image

and was immediately disappointed.

AIPRM for “Potential Risk Managment” :rofl::rofl::rofl:

It scraped up 1 super spammy site that published some basic ChatGPT output, with totally made up details and the result was poor.

Then I realized

To be fair, my prompt here sucks.

and it did suck.

Like all those quickly entered 1-sentence prompts do.

After some prompt engineering and lots of testing I got something quite decent

That used multiple, trusted, sources.

Here’s the full monthy it gave me - first shot with my new “Better Browsing” prompt,
and I think it is very decent and very accurate.

What are your experiences with ChatGPT using “GPT4 with Browsing via Bing”?

If you’ve not enabled it yet, some users need to enable beta featurs in the “Settings”

image

1 Like

How’s that possible. I thought it only searches bing.

It only did search Bing, and then interpreted trustworthy results it found

1 Like

Does this mean that the right to interpret the “truth” will return to the hands of those “authorities”, oh, I seem to have seen “indulgence notes”, a feeling of the dark middle ages. I think we need to establish a transparent and open monitoring mechanism to prevent the people’s right to know from being fooled by the media. There are always some “idiots” in the world to realize their value in this world by doing evil: always reminding people that there is still an evil side in this world. I think “multi-party proof” will be an effective way. I never expected that one of the most profitable industries in the future would be the news industry that has been silent for many years, so go buy stocks of media giants such as Murdoch, and I guarantee that it will double in the future! :sunglasses:

2 Likes

I think we can establish the concept of “blockchain” in the information field, that is, there are witnesses in the area where the incident occurred to provide relevant falsified evidence. Photograph the evidence and send it to exchanges such as our information “Ethereum” for confirmation. If the information is true, we will sue the counterfeiters and distribute the fees obtained from the prosecution to those who provide real evidence. I think this is a wonderful idea. The information authenticity guarantee measures, which maximize the information authenticity in the hands of the people, not the hands of the news giants, oh, I seem to have found another opportunity to short when Murdoch’s stock is trading at the daily limit! If I were on Wall Street now, I should be richer than Musk! Hahaha. But I care more about the overall future of humanity than money, mainly because the thought of the Dark Ages gives me a headache. :smiling_face_with_tear:

Some interesting thoughts @Alfgogering

The selection of more than 1 source is the first goal.

That original dumb 1 liner only used a single source, and that was rubbish, to begin with. Upon review it immediately stood out as AI generated spam.

Search engines have the problem with spam, and lesser with fake news for over two decades.

Solving the trust problem works via links from other trusted sources. It is what you describe in a different way by “people taling pictures”…

…human votes. However nothing is perfect and so those votes (links) are gamed for 20 years now.

The recent suggestion by OpenAI on how to steer AI certainly also touches on this topic.

It’s developing nicely, using the new Prompt-Dropdown-Variable feature @RealityMoez

1 Like

The way you use it with plugins… WOW.

So, Does this mean: based on the drop down choices, the prompt changes? and ChatGPT can’t see these settings or what?

Yes, indeed.

ChatGPT can’t see these settings or what?

Correct. The Prompt Template and the options (“VARIABLE”) are merged as usual

2 Likes

The prompt should have been published today…

…then ChatGPT released “some improvements”

Tell us the bit about “avoiding issues” again, Christoph :rofl:

Yes, yesterday I left a message under Sam’s Twitter about my plan: to realize digital democracy by establishing a credit “blockchain” system, which is similar to when it is discovered that rumors are spread through AI technology, and by sending Credit “Blockchain” users issue rewards (of course, users who receive rewards are also divided into credit ratings, which are similar to the weights in the current virtual currency. When encountering particularly difficult events, then we will send “masters” to investigate the integrity of the event The process, believe me, the user who can get the credit rating of “Master” at that time is definitely not an idle person, at least he does not work for money), let them collect evidence to verify the authenticity of the event, if the event is fake, then We can prosecute the liar through the chatGPT lawyer, and then send the compensation paid by the liar to the “detective”. Of course, the credit exchange can draw a certain handling fee (although I think this handling fee should be paid by the United Nations), and then, We have obtained a “social credit model” in the AI era that can operate efficiently. I mainly refer to the decentralization of “blockchain” technology, so that the truth can be in the hands of the people as much as possible, and capital can be used as an auxiliary. To make this mechanism run for a long time under the current system, I know that it has the taste of an AI trial court, but the difference is that there can be more “opportunities” under this mechanism, which are similar to our fish at the beginning. The ancestors chose to climb onto the land instead of giving the initiative to such news giants as Murdoch, and let the politicians behind these news giants control the truth, although the latter AI tribunal will also be understood by the politicians Rules, but as I said, those users who can become “masters” should have an unyielding “Sherlock Holmes” in their hearts. :partying_face:

It’s coming - here’s a teaser video

1 Like

Really well done, I’m impressed :clap:

1 Like

You can try “developer mode”, of course, this requires you to find the corresponding “spell” on the Internet. :thinking:

1 Like

It is here now

2 Likes