AI Safety Summit: Sufficient or Superficial?

The AI Safety Summit took place in the UK last week.

The main outcome is the signing of an international statement on AI safety, "an unenforceable wish list of general ideas with no teeth" according to Chris Penn, the first speaker I ever followed on AI, back in 2017.? ?

Drill through the reporting (Media, Twitter, TikTok) and it's all mainly:

• That the event occurred (=massive PR coup for the British government)
• People in awe of being at Bletchley Park / in the same room as Elon Musk.
• "It won't take your jobs!"

In the same week, the US Government came out with an Executive Order on AI security which is already out of date. See this video from Chris Penn on exactly why this is so. This bit especially on misuse of AI and cybersecurity.

Elon Musk isn't a huge fan of OpenAI. If you were cynical you might say that he agreed to headline the AI Safety Summit in order to divert attention from OpenAI's Dev day which took place yesterday and was like an early Christmas gift from the parents of ChatGPT. More on that later.

Elon Musk's statements on the need to protect public safety ring true and must be listened to, even if he is the guy that killed Twitter. AI is the most disruptive force in history.?We must take it seriously.

Interestingly he advocates reading Iain M. Banks' brilliant sci fi novels to understand the potential future we are facing. Mad that, as I had the same thought myself last week and paid a visit to my local bookshop in Athlone to buy some copies, only to discover that they'd sold out of Iain M Banks.

I did some digging around and found this report that was presented on day 1 on the capabilities and risks. I got Claude 2 to summarise it for me:

??? The report cautions that rapid advances in AI capabilities could outpace efforts to ensure safety and beneficial impacts. (Yeah, like the way it has already with the US Executive Order in one week? How exactly are our governments going to test? models? Anyone else watch the Facebook or TikTok Congressional hearings and cringe at the questions being asked?)?

?? The report warns of risks from criminal or state exploitation of AI to automate malicious activities like cyberattacks. (BBC's Lazarus Heist is a good place to learn more about that).

?? Looking longer-term, there are concerns that AI systems may take actions that reduce human control, if not properly aligned with human values. (This is one is a humdinger and the one that worries me the most. The sooner we move ourselves on from the 'is AI going to take our jobs' question to a more heightened sense that this is an emergency that is coming soon and needs to be planned for, the better.

The best description of what we are facing is from Mo Gawdatt in a brilliant and must-watch episode on Steven Bartlett's "Diary of a CEO" in which the ex-GoogleX Officer, describes it like an old fashioned game of tetris.. how there's always that one piece, which screws up the rest of the game, if you don't place it correctly, the game gets more difficult.?

We are in that stage now - where we're facing lots of pieces coming at us and we absolutely must make the right decisions.? To play this game of Tetris, what skills will you need and what type of provider is needed? A really good place to start is that Mo Gawlam + Steven Bartlett interview. It's long but he's very understandable and I promise, he doesn't talk in tech! Click the pics below to take you there.?? ????

Something I noticed, he mentioned (at 0.94) that if you're in an industry that is threatened by AI, you've got to get over your fears and learn how to utilise AI in your work. I would suggest you start now by making a list of all the tasks you do in a day. You might find many of them can be automated or done quicker better faster using AI. You don't have to worry about which ones, how, and what's the best way, because that's where I can help. Check out our list of AI courses for everyone from non-technical, to marketers, to specialists.? There are no places left for this year, but we are booking for 2024.?

The threat to jobs risk isn't going to happen in the immediate short term. But you must master AI if you are going to survive and compete.

That's it for now. Be back when I have more news.