Ryan Gibbons’ Post

Hi all, I found something cool! There is a great effort by the Stanford Human-Centered Artificial Intelligence (HAI) center to track implementation of Biden's executive order 14410 - the AI one released to much fanfare last year. We are 90 days in and some deadlines are past. So how are they doing? Better than expected in terms of meeting deadlines, but it is difficult at this point to tell if they are doing a good job at any of the deliverables. Still, if you want to keep an eye on how things are progressing and track the orders line by line then the HAI tracker resource is amazing. I was about to do some basic analysis of the efforts when a colleague showed me this great tool. If you are in a critical infrastructure sector or feel that some of these efforts might affect you (Hint: if you are planning to use AI, then they will), I'd recommend you give it a look. https://lnkd.in/e2pNpgY3

Just finished the course “Introduction to Artificial Intelligence”! #artificialintelligenceforbusiness

Favourite meme of the day, posted by Elon Musk in Twitter as commentary on the recent UK Safety Summit. So true! Original creator unknown.

Twenty-eight nations came together to discuss AI safety at Bletchley Park in the UK, famous for the WWII codebreakers. There the countries agreed to cooperate on building AI capabilities that are 'human-centric, trustworthy, and responsible.' The focus was on understanding risk (naturally) and there was a commitment to identify "...AI safety risks of shared concern, building a shared scientific and evidence-based understanding of these risks, and sustaining that understanding as capabilities continue to increase..." But my favourite part was the commitment to build and provide testing tools, which is one of the trickier parts of AI governance. From the document - "This includes, ... appropriate evaluation metrics, tools for safety testing..." I'm sure there will be some heated debates around what the end result looks like, but at least it is one of the main objectives. Overall, while is encouraging to observe that the text frequently refers to ongoing plans to cooperate, I'm not convinced that anyone would let this impede their own efforts at this point. Still, understanding risk is a hopeful first step and these types of events can leave a lasting legacy if the participants are committed. https://lnkd.in/efNUBNM4

The EU AI draft legislation is close to becoming law. On the positive side, I appreciate the transparent and deliberative nature of the legislative process, having watched this since the draft was released earlier this year. I also like how they have structured the document into clear levels of regulation that are associated with risk - from prohibited use cases all the way down to not applicable. On the other hand, I have concerns about holding developers liable for their products in the way prescribed. For defects sure - similar to homes and cars - but this feels more like making Microsoft liable for anything written in MS Word. I get that they are trying to take a strong stand to regulate AI, but this may end up only limiting innovation and use of new technologies by law abiding companies. Any other key pro/cons that you see? https://lnkd.in/ezC79y3b

Just finished a really interesting week at Gartner IT/Xpo 2023 in my new hometown. A few takeaways that might be helpful for others. AI AI AI AI AI Gartner is not concerned about existential risk or AGI and see it as a key value add across all verticals over the next few years. Most CEO's apparently have the same take according to their polling. Data Strategy Payoff time for those that have invested in inventory and data management competencies. AI efforts struggle and don't make it into production without data, plans, and resources. No shortcuts here, so may as well get started. Productivity This is where most AI efforts will focus on the beginning, and rightly so. It takes time to build up the organisational skill and capabilities to tackle the more advanced use cases. Plus, they can be cost neutral (or better) which really helps in cost controlled settings. Security and Governance AI Governance is about customer and stakeholder trust in how you operate, ultimately. Even while waiting for the regulators and industry to figure things out, start implementing security and governance to manage risk and let people know that this aspect is important. Document decisions and have well understood responsibilities. Much more to come here but that is a great place to start. There were some other gems as well (API, workforce engagement, etc), but AI dominated. How are things going for you on this front?

Recommended for anyone wanting to get started in the code security space or to validate their approach. Of course there are tons of other tool options, but the ones included are a great starting place.