Hello and welcome to GG News!

In this edition we will obviously cover the ongoing OpenAI saga and the Sam Altman, Greg Brockman Hokey Pokey!

We also discuss our recent Raspberry Pi project that enabled us to remotely connect to our old intruder alarm system giving us the functionality to set and unset the alarm and receive alarm activation notifications by sending simple Telegram messages from anywhere in the world!

For all this, as well as the latest tech news and information on the most useful AI tools keep reading and please share with your friends and colleagues!

Please remember to check out our affiliate partners and use our links to sign up for their services. You won’t pay anything extra, but we will receive a small commission to help us keep this newsletter free for you all to read!

OpenAI: Fired, Re-hired, Chaos transpired, Microsoft inspired?

Okay now I’m sure we are all up to date on the recent OpenAI saga but let’s run through a quick summary.

November 17:

• OpenAI dismisses CEO Sam Altman due to communication inconsistencies with the board.

• Mira Murati, OpenAI’s CTO, becomes the interim CEO.

November 17-18:

• Microsoft’s Satya Nadella reaffirms commitment to OpenAI post-Altman's exit.

November 17-18:

• President Greg Brockman resigns in support of Altman.

November 18:

• Industry figures express support for Altman on social media.

November 18:

• Reports emerge citing disagreements between Altman and the board, particularly with Ilya Sutskever, over AI safety and company direction.

November 18:

• Altman expresses gratitude for support received.

November 19:

• Investors urge OpenAI to reconsider Altman's dismissal.

November 19:

• Fears arise of employee resignations if the board doesn't step down.

November 20:

• Altman and Brockman not returning to OpenAI - confirmed.

November 20:

• Microsoft announces Altman, Brockman, and colleagues will lead their new AI research team.

November 20:

• Sutskever expresses regret for his part in the events.

November 20:

• 505 OpenAI employees threaten to resign en masse if the board doesn't step down.

November 21:

• Altman communicates unity and commitment towards OpenAI’s future.

November 21:

• Reports suggest investors considering legal action against OpenAI.

November 22:

• OpenAI agrees in principle to reinstate Altman as CEO, revamps the board.

November 22:

• Altman expresses love for OpenAI, intent on keeping the team together.

November 22:

• Microsoft praises changes to the OpenAI board and looks forward to continued partnership.

November 22:

• Shear and Brockman express satisfaction with the resolution.

Pretty crazy huh?

 “Risk to humanity” rumoured as cause of OpenAI turmoil. ☠⚡💀💀

Reports suggest that the reason behind Sam Altman’s original firing from OpenAI may be connected to reports of internal communication sent to the OpenAI board prior to his departure.

Sources from within OpenAI have apparently suggested that this internal communication suggested a genuine concern that a secret internal project known as Q* (pronounced “Q star”) had given researchers concerns that the project may have demonstrated super-intelligence.

How this is linked to Altman’s firing is still unclear but there have been previous reports of friction within OpenAI regarding the pace at which OpenAI is progressing, with some elements within the company allegedly suggesting that Altman is pushing too hard without the necessary checks and balances in place.

What have we been having fun with? 🎉

So, this week was actually pretty interesting. We decided that we would update the GG News intruder alarm with a Raspberry Pi4B that we have had for a while.

We have a rather old, basic alarm system that works perfectly but we wanted to add the ability to remotely arm and disarm the alarm, and to get notifications when the alarm activated.

If you look online there are a number of projects that can help with this but none that really suited what we wanted and none of the team can write code!

So, what did we do? We asked ChatGPT of course!!

Now, full disclosure this was definitely not a case of asking ChatGPT to write us some code and then dropping it into the Pi and there you go! But it also wasn’t a complete nightmare.

First, let’s lay out in a little more detail what we were trying to accomplish:

The idea was to somehow be able to set and unset the alarm and to also get notifications if the alarm activated and we wanted to be able to do this from anywhere in the world! 🌍

To handle the triggers and receive the notifications we decided to use the Telegram app. There are a few reasons for this but mainly it was because it’s free and very easy to integrate with. Now the whole process is a bit long-winded for this newsletter, but you can access the code we used for the Pi here and details for setting up a Telegram bot and integration can be found here.

Please play around with the code by downloading and installing Thonny. Pro tip if you want help understanding how the code works try the following prompt in ChatGPT:

  “Please can you look at the following code for a raspberry pi and explain it to me like I am a 10-year-old. Please explain the code line by line and answer any follow up questions I may have. Then using the code as a reference please ask me 10 questions to ensure I fully understand.” Insert code here.

You will need to alter the code section shown below to input your API token and Group ID for Telegram as explained in the second link above.

# Telegram Bot Token
TOKEN = "INSERT API TOKEN HERE"
# Telegram Group ID
GROUP_ID = INSERT TELEGRAM GROUP ID HERE

Having the code installed on a Raspberry Pi should give you plenty to play with and should teach you a lot, but if you have no experience at all with using a Raspberry Pi then you will need to look for some online resources to help with the nitty gritty. Alternatively, if you would like further help then we could possibly put together a full “how to” video if we get enough requests!

Another point worth mentioning is that using ChatGPT to write code can get frustrating. Often, we would get errors in the code which we would feed back to ChatGPT. ChatGPT would then fix the error only to break something else! We would then get stuck in a frustrating loop for two hours of fixing one issue while breaking something we had previously fixed.

We found two solutions to this that really helped. One was to build the project in stages getting one thing at a time to work e.g. set up the input to receive alarm activations on GPIO 17. Once this was working, we would start a new conversation, input the code and ask ChatGPT if it understood the code, and then ask it to modify the code with the next stage added. Alternatively, we would feed the code into another language model such as Claude or one of the trained bots on poe.com. We seemed to have the most success using forefront.ai.

The end result was an alarm that could now be fully controlled using the Telegram app. If you look at the code, you will see that we ended up with the following capabilities:

If we type “1” in the assigned Telegram group chat, then the alarm is set/unset (this toggles).

If we type “P” then the panic input is triggered on the alarm.

The above two were connected to the alarm via a double relay module we purchased here. Obviously, the connections and set up will depend upon which alarm you have.

If we type “Reboot” the Pi will reboot, this was actually something that ChatGPT added by itself which was quite useful during our initial testing as at some point in the development we were having problems with the code that would cause the Pi to lock up and because the Pi was installed near the alarm unit it wasn’t that accessible. Being able to remotely reboot the Pi was amazing.

Above is a screenshot of the Telegram setup in use.

As mentioned, if you like this and want more information or are struggling to follow along, let us know and if there is enough interest, we will put a quick video together.

What’s in the news? 📰

Sam Altman - Out, no in, hang on out, no he’s in...yay🎉 we think?😕🕺🏻

Anthropic ups the ante - Anthropic has released Claude 2.1 with an increased token limit of 200000 tokens. 📈

Bard answers questions on YouTube videos - Googles Brad now has the ability to directly answer questions on YouTube videos 👀

Nvidia being sued after video call exposes stolen data! - Nvidia employee accidently reveals data he stole from his previous employer 💣💻

Binance CEO admits money laundering - The CEO of finance company Binance Changpeng Zhao has stepped down after admitting money laundering violations. 💸🧺

British national library confirms employee data breach after ransomware attack - The British National Library is still struggling to recover from an October ransomware attack by the Rhysida ransomware gang.

Omegle closed down after out of court settlement - "I feel personal pride that no more children will be added to Omegle's body count," says the woman who successfully forced the infamous chat site to shut down.

AI tools

1. Vidnoz - AI video creation with a template library and lots of avatars.

2. HooopsAI - Get trading analysis for investment strategy.

3. Learning studio AI - Ai course creation tool.

4. SummaristAI - Summarise any book by just entering the title.

5. InSummary - Summarises your work activities automatically for your wider team.

Affiliate links

Below are our Affiliate links to sign up for Pabbly and voicemyai:

Pabbly has a special one-time payment plan which gives lifetime access for a single one-off payment, you can see these one-time payment options here.

For those of you who would like to look at their monthly recurring plans please check out the link here.

This is our affiliate link to sign up for voicemyai😀 

This week’s quote:

“You must always be prepared to place a bet on yourself, on your future, by heading in a direction that others seem to fear.”

-Robert Greene