Apple is FINALLY Talking About AI

0:01Today on the AI Breakdown, we're previewing the UK's AI Safety Summit, which kicks off today. Before that on The Brief, is Apple finally getting into the AI race? The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown.net for more information about our YouTube channel, our Discord, and our newsletter.

0:24Welcome back to the AI Breakdown Brief, all the AI headline news you need in around five minutes or less. Well, friends, an unexpected thing happened yesterday. Apple finally showed up to the AI wars. One of the big questions ever since ChatGPT launched, really, but especially since companies including Microsoft and Google and Amazon all absolutely went into a flurry to try to stake their claim around artificial intelligence, has been when Apple was going to get in the game and what they were going to do. Recently, it seems as though there has been more action on that front. We've had reports that they're spending millions of dollars a day training their own models, and more recently we heard that the next iOS operating system to come out next fall is likely to be AI integrated back to front.

1:08Well, yesterday at 8pm Eastern, Apple held a new product event where they announced the latest in their line of iMacs and MacBook Pros. This is notable for one big reason, and that reason is the M3 chip. The M3 is the latest in Apple's custom chip line, and in their presentation about it, AI was a term that they actually used. AI YouTuber Matt Wolf wrote, Damn, my video editing workflow may be moving over to Mac pretty soon. And they've actually said AI at least twice in their presentation. Never expected to hear that from Apple. Ye Ding wrote, For those of us in the AI space, this was the most interesting part of Apple's M3 announcement yesterday.

1:46You can now run the biggest open source LLM, Falcon, with 180 billion parameters, with low quality loss on a 14-inch laptop. What a time to be alive. Indeed, Inc.'s summary of the event focused in on this topic as well, saying, with the M3, Apple is finally talking about AI. During the event, for example, they said, with the M3, the neural engine is up to 60 % faster than in the M1 family of chips, making AI ML workflows even faster while keeping data on device to preserve privacy. The Inc. piece goes on. Apple even went as far as to highlight that the increased memory capacity on the M3 Macs supports, quote, workflows previously not possible on a laptop, such as AI developers working with even larger transformer models with billions of parameters.

2:27As Inc. sums up, Apple isn't just talking about the features it builds into its own products that take advantage of its own capabilities, but is also explicitly positioning its new MacBook Pros as a tool for developers building AI products. That's a pretty significant change. Now, this also validates what people have been assuming about part of what Apple's AI strategy would be, which is effectively racing to have the hardware catch up to a point where Apple doesn't have to compromise on its long-held positions around running applications on device rather than having to rely on the cloud and focusing on user privacy.

2:59My commitment to you, the AI Breakdown listeners, is that I will, for the sake of science, buy the most powerful M3 Max MacBook Pro model and tell you how it works. Next up, a company that was in the news recently for raising a bunch of money is now back in the news for raising even more money. The information writes, Mistral, a wannabe open AI of Europe, seeks$300 million. Now, this is a company that raised a$113 million seed round over the summer, four months ago to be exact. Now the roster of talent included former Meta and Alphabet researchers, and more than that, the open-source Mistral 7b model that they put out has quickly become a favorite of developers.

3:35The information says that the round is expected to value the company at over a billion dollars, pre-money. The information also writes, Mistral is also developing its products in line with stricter European regulations like the European Union's AI Act, and has emphasized privacy and security according to its pitch deck. The information speculates that could help it beat out OpenAI and other companies like it when it comes to attracting European corporate customers. Now, while developers have loved Mistral, there has been some controversy around its open source approach, which lacks some of the safety guardrails that other closed models include.

4:05If you want to read their deck, which is actually a seven-page strategic memo, I will include that in the show notes. Now, moving to our next story, one of the things that we have talked about as a key theme for this fall is the integration of AI into the workflows and experiences that we already have. Well, as this story suggests, sometimes that integration is going to be a little bit bumpy. When The Guardian published an article about a woman found dead at a school in Australia, Microsoft's news aggregator Microsoft Start created an AI-generated poll designed to increase user engagement with the fairly distasteful question, what do you think is the reason behind this woman's death?

4:40Murder, accident, or suicide? Now, of course, Microsoft pulled the poll, but there was some serious reputational harm already caused. The Guardian wrote a scathing letter to Microsoft president Brad Smith, Microsoft has so far not commented. Over in China, Chinese tech giants continue to release upgraded AI models to keep the pressure on US rivals. Alibaba announced the 2.0 version of their model as well as a Gen.ai service platform which is designed to help companies build their own generative AI applications using their own data. In this way, the company is competing with both Azure as well as AWS's Bedrock.

5:13Next up, as earnings season continues on, market watchers have a close eye on AMD. AMD has anticipated to release its MI300 chip this quarter, and the product is seen as key in AMD's battle to try to keep pace with market leader NVIDIA. AMD's shares were up 2.1 % in the hours before the call and are up 47 % on the year overall. Now speaking of NVIDIA, a little inside baseball from that company, apparently they are piloting a generative AI specifically for their engineers called Chip Nemo. In what is something of a dogfooding moment, the company's CTO in a keynote on Monday talked about how they have been testing a specific LLM to boost the productivity of their chip designers.

5:51The model is custom designed and trained on NVIDIA data with three different purposes in mind, as a chatbot, as an EDA tool scriptwriter, and as a summarizer of bug reports. Another AI integration experiment comes from Siemens and Microsoft, who are collaborating on something called the Siemens Industrial Copilot Scheme. The idea is to bring generative AI to industries including manufacturing, transportation and healthcare. Reuters writes, the project will create AI co-pilots to assist staff at customer companies as they design new products and organize production and maintenance. It examines information gathered by Siemens and helps customers quickly create, improve, and debug complex automation codes and shorten simulation times at their factories and other facilities.

6:30So there you have it, friends. Tons going on. Lots of big company engagement. Lots of startup action. It is a very fast-moving fall here in the world of AI. And I appreciate you hanging out here at the AI Breakdown to try to keep track of it all. That's going to do it for the AI Breakdown Brief. Next up, the main AI breakdown. And now, a word from today's sponsor. Are you interested in how two top-of-mind trends, AI and crypto, can work together? If so, I have the perfect podcast recommendation for you. Web3 with A16Z Crypto, the chart-topping show brought to you by venture firm Andreessen Horowitz.

7:06Web3 with A16Z Crypto is your definitive resource for the future of the internet, whether you're already building in these spaces or simply curious about what's next. If you need a place to start, they recently released an excellent episode with Stanford cryptography professor Dan Bonet and former Google Xer Ali Yahya in conversation with host Sonal Choksi about the intersection of AI and crypto. From fighting deepfakes and proving humanity to large language models like ChatGPT, they cover it all. I highly recommend checking it out, especially if you'd like to learn more about how AI and crypto will impact our everyday lives.

7:38Beyond Crypto and AI, this show is for creators seeking more ways to truly own their work, for business leaders trying to prepare for the future today, and for innovators exploring trending tech topics. So go ahead, listen to Web3 with A16Z Crypto wherever you get your podcasts. Hey guys, before we get into the main part of the episode, I wanted to put out a call for a few different types of people I am trying to hire right now. The TLDR is that I have a pretty exciting AI education-related project that I'm not really ready to share more details of just yet, but which suffice it to say, I am incredibly excited about.

8:13On that front, I am looking for two types of people. The first is developers and UI slash UX designers, and the second is content producers, basically technical and non-technical builders. If you are either of those things and you want to learn more about what we're building, send me a note at jobs at breakdown.network, and please share examples of what you've built or created. So that can be websites or apps in the case of developers or UI UX designers, or it can be content that you've produced in the case of content creators. Again, that's jobs at breakdown.network. Looking forward to hearing from you.

8:48And now on with the show. Welcome back to the AI breakdown. This morning, the UK's much anticipated AI safety summit began. This is one of the largest and most high-profile events to tackle head-on the issues of AI safety and X-risk, and how governments might be involved not just in the immediate challenges of AI, such as bias and hallucination, but also in thinking about those big, society-altering types of issues. Now, this event has seen a huge amount of controversy. There has been controversy around the British government's decision to invite the Chinese government to participate. There has been controversy around the very focus on these most extreme issues of AI safety.

9:27There's been controversy around who was going to attend and whether world leaders were snubbing the Rishi Sunak government. But to understand how they are thinking, let's start by reading a very short op-ed that the Prime Minister Rishi Sunak himself published just a couple days ago on Monday called The World Must Act Now on AI. The PM writes, I believe nothing in our foreseeable future will transform our lives more than artificial intelligence. Like the coming of electricity or the birth of the internet, it will bring new knowledge, new opportunities for economic growth, new advances in human capability, and the chance to solve global problems we once thought beyond us.

10:02AI can help solve world hunger by preventing crop failures and making it cheaper and easier to grow food. It can help accelerate the transition to net zero. And it is already making extraordinary breakthroughs in health and medicine, aiding us in the search for new dementia treatments and vaccines for cancer. But, like previous waves of technology, AI also brings new dangers and new fears. So if we want our children and grandchildren to benefit from all the opportunities of AI, we must act and act now to give people peace of mind about the risks. What are those risks? For the first time, the British government has taken the highly unusual step of publishing our analysis, including an assessment by the UK intelligence community.

10:40As Prime Minister, I felt this was an important contribution the UK could make to help the world have a more informed and open conversation. Our reports provide a stark warning. AI could be used for harm by criminals or terrorist groups. The risk of cyber attacks, disinformation, or fraud pose a real threat to society. And in the unlikely but extreme cases, some experts think that there is even the risk that humanity could lose control of AI completely, through the kind of AI sometimes referred to as superintelligence. We should not be alarmist about this. There is a very real debate happening and some experts think it will never happen.

11:11But even if the very worst risks are unlikely to happen, they would be incredibly serious if they do. So leaders around the world, no matter our differences on other issues, have a responsibility to recognize those risks, come together and act. Not least because many of the loudest warnings about AI have come from the people building this technology themselves. And because the pace of change in AI is simply breathtaking, every new wave will become more advanced, better trained, with better chips and more computing power. So what should we do? First, governments do have a role. The UK has just announced the first ever AI Safety Institute.

11:43Our institute will bring together some of the most respected and knowledgeable people in the world. They will carefully examine, evaluate, and test new types of AI so that we understand what they can do, and we will share those conclusions with other countries and companies to help keep AI safe for everyone. But AI does not respect borders. No country can make AI safe on its own. So our second step must be to increase international cooperation. That starts this week at the first-ever Global AI Safety Summit, which I'm proud the UK is hosting, and I'm very much looking forward to hearing the Indian government's views.

12:13What do we want to achieve at this week's summit? I want us to agree the first-ever international statement about the risks from AI, because right now we don't have a shared understanding of the risks we face. And without that, we cannot work together to address them. I'm also proposing that we establish a truly global expert panel, nominated by those attending the summit, to publish a state of AI science report. Over the longer term, my vision is for a truly international approach to safety, where we collaborate with partners to ensure AI systems are safe before they are released. None of that will be easy to achieve, but leaders have a responsibility to do the right thing, to be honest about the risks, and to take the right long-term decisions to earn people's trust, giving peace of mind that we will keep you safe.

12:50And if we can do that, if we can get this right, then the opportunities of AI are extraordinary, and we can look to the future with optimism and hope. So I think there are a couple things that are worth noting about this. First is that, especially compared to the Americans, the UK government is definitely much more comfortable talking about actual X-risk and actually losing control of AI. Now, they obviously couch it here and say there's a debate happening with some experts thinking it will never happen, but that doesn't change the fact that they are literally and figuratively putting it on the agenda.

13:18The second thing that's interesting is what they're actually trying to achieve at the event. On the one hand, I think one could make an argument that simply bringing people together to talk about these issues face-to-face in the context of an extremely influential and powerful group of people is valuable in and of its own right. However, the UK government is not just trying to be a host of a good conversation. They want to be the leader in a new type of international action. However, given who is coming to this event and how things have evolved, their ambitions for what that action can be are clearly a little bit tempered now.

13:49We just heard what Prime Minister Sunak said, I want us to agree to the first ever international statement about the risks from AI. I want to establish an expert global panel to publish a state of AI science report. And over the long term, I want to collaborate to ensure AI systems are safe before they are released. Now, of course, what this event and these would-be actions are competing with are A, the individual actions of nations like the United States, which obviously got out ahead of this event by releasing the Biden administration's something like 110-page-long executive order on AI earlier this week, and on the other hand, other ongoing processes around global AI regulation.

14:24One of those is the so-called G7 Hiroshima process, and that is clearly making progress in its own right. On Sunday, Reuters reported that the G7 was about to agree to an AI code of conduct for companies. This was designed to be a voluntary code of conduct that would create a set of benchmarks for how major countries govern AI. The G7 documents seen by Reuters show that this was an 11-point code that, quote, aims to promote safe, secure, and trustworthy AI and will provide voluntary guidance for actions by organizations developing the most advanced AI systems, including the most advanced foundation models and generative AI systems.

14:56Now, a lot of this sounds like some of what we saw in the Biden administration executive order as well. Reuters writes, the code urges companies to take appropriate measures to identify, evaluate, and mitigate risks across the AI lifecycle, as well as tackle incidents and patterns of misuse after AI products have been placed on the market. Now, back to the AI Safety Summit that's happening in the UK today and tomorrow, let's talk about what's on their agenda. The UK government released a PDF called Introduction to the AI Safety Summit. In it, they echo many of the same themes from Rishi's letter that we just read, namely that there are lots of opportunities but also serious risks that we need to tackle.

15:29Now, one thing that's worth noting is in the section called Scope of the AI Safety Summit. What is Frontier AI? They write, the summit will focus on certain types of AI systems based on the risks they may pose. These risks could stem from the most potentially dangerous capabilities of AI, which we understand to be both at the frontier of general purpose AI, as well as in some cases, specific narrow AI, which can hold potentially dangerous capabilities. So the scope of the summit then, as they show in a nice little chart, is one, narrow AI with dangerous capabilities, e.g. AI models for bioengineering, and two, Frontier AI, e.g.

16:02cutting-edge LLMs. Within that, they are focusing on two particular categories of risk. One, misuse risk, which they describe as where a bad actor is aided by new AI capabilities in biological or cyber attacks, development of dangerous technologies, or critical system interference. The second is loss of control risks that they say could emerge from advanced systems that we would seek to be aligned with our values and intentions. Now, why wouldn't they address the more immediate risks? They actually talk about that in this paper as well. They say, this is not to minimize the wider societal risks that AI, both at the frontier or not, can have, including misinformation, bias and discrimination, and the potential for mass automation.

16:39However, they write, the UK considers that these are best addressed by the existing international processes that are underway, as well as nations' respective domestic processes. For example, in the UK, these risks are being considered through the work announced in the White Paper on AI Regulation and through wider work across government. It is therefore important that the focus of the summit is complementary and not duplicative of these existing efforts, and that the UK and other nations continue to work at pace across all these forums to address the full range of risks. So what's really clear from this, even clearer than in some of the other places that they've written the purpose of this event, is that this is meant to fill in the gap around the biggest risks, the ones that have serious potential for loss of life.

17:18It feels like there is an argument underlying that last little piece that we just read, that those issues are not currently being addressed in a meaningful way, and that that gap is a dangerous one for humanity to leave unfilled. Now, when it comes to how much impact it's likely to have, there has been some challenge in securing the participation of world leaders. However, the event had a coup when earlier this week it was announced that Elon Musk was expected to attend the event in person, and then subsequently would be hosting Prime Minister Rishi Sunak for a conversation on Twitter Spaces. Now, Elon is nothing if not a controversial figure.

17:50But he is also a media lightning rod, which means that simply through his presence, this event is likely to have much more attention paid to it, which I think is precisely the point. Beyond Elon, the government of the UK also published the governments and organizations that were confirmed as attendees, although without giving a lot of detail about who from those organizations or governments was actually going to come. From academia and civil society, you have a huge range of people, including the Usual Suspects, Future of Life Institute, Open Philanthropy, Stanford, Berkeley, Oxford, and more.

18:20Governments, you have Australia, Brazil, Canada, China, France, Germany, India, Indonesia, Ireland, Israel, Italy, Japan, Kenya, Saudi Arabia, Netherlands, New Zealand, Nigeria, Korea, Philippines, Rwanda, Singapore, Spain, Switzerland, Turkey, Ukraine, UAE, and the US of A. And then some of the notable industry and related organizations include Amazon Web Services, Alibaba, Anthropic, Cohear, Databricks, Google DeepMind, Google, HuggingFace, IBM, Inflection AI, Meta, Microsoft, Mistral, NVIDIA, OpenAI, Palantir, Salesforce, ScaleAI, StabilityAI, and of course, Elon's XAI. Lastly are the multilateral organizations, including the European Commission, UNESCO, United Nations, and the Council of Europe.

18:58China, for its part, is sending a Vice Minister of Science and Technology. So this is going to be interesting to see. What I'm watching for is to understand whether this will be a really great conversation where some important people talk about and agree on the scope of some risks, or whether there will actually be pathways to action beyond that. I don't know exactly what I would consider success, but if nothing else, I'll be paying attention to see how the world and how the media specifically engages with the event. Can the presence of someone like Elon make it a thing that everyone's talking about, or will it just fade quietly into the background?

19:30That is what we will watch, and I appreciate you watching it along with us here at the AI Breakdown. Until next time, peace! Thank you.