AI Safety Summit – CEPA

Ronan Murphy
So good afternoon, all people. Thanks very a lot for becoming a member of this name. This can be a press briefing hosted by the Middle for European Coverage Evaluation, and we’re going to be discussing points in and across the upcoming AI Safety Summit and what may come from it, what won’t. And we’re taking views from either side of the Atlantic right here right this moment. We’ve received, joined by senior fellows from CEPA, one in Brussels, Invoice Echikson, who has a few years expertise in know-how, each on the within and reporting from the skin, and can also be the editor of bandwidth, our on-line journal on know-how coverage. And in Washington, DC, we’ve received CEPA Senior fellow, Heather West once more, a few years expertise working throughout the know-how enterprise and consulting and dealing within the authorized discipline, knowledgeable in AI coverage, cyber safety, and we’re very glad to have each of you with us right here right this moment. Simply to talk rapidly to fundamental housekeeping, we’re on the report, and we can be offering a recording and a transcript after the decision. And my colleagues from from CEPA, together with Michael, who you’ll be able to see within the prime nook, most likely, are on the decision and can fill in on any of the element concerning that on the finish of the decision. So please, you’ll be able to ask questions at any stage. And by way of the format, we’ll, I’ll throw a few opening inquiries to our to our visitor audio system, however we actually encourage, I actively encourage you so as to add questions by the chat. If you happen to do, I’ll most likely ask you to learn them out your self, for the report, for everyone, however if you happen to’re not able to take action, I’ll learn them out in your behalf. That’s no downside. And, and, yeah, with that in thoughts, I believe we will, we will get going. And it’s a, AI is a subject you’ll be able to’t keep away from, and for for many individuals on this enterprise, it’s introduced up daily in all places. And why we’re having this name is that there’s a summit subsequent week in San Francisco, the AI Safety Summit, stemming from efforts began by the UK to attempt to put some framework, some worldwide framework, on an strategy to AI security particularly, and what the, what the bottom is for that everybody can agree on. So I would ask, Heather, if you happen to perhaps wish to set the scene as to what we what we would anticipate and what is likely to be good, then each aren’t essentially the identical factor from the gathering,

Heather West
Positive. So, so I believe it, this can be a, I consider, the primary assembly of this worldwide community of AI Safety Institutes and this convening is actually supposed to create an atmosphere for world collaboration, which is actually great. And so they’re bringing collectively very particularly technical consultants, they usually’re bringing people collectively who actually know the right way to do the work and perceive the know-how which actually provides us a chance to maneuver the work ahead. Wanting in direction of that, that very technical measurement and science based mostly guidelines of the street for AI security, the way in which that was specified by the Seoul Assertion, must be actually attention-grabbing. And so they they’ve received numerous international locations becoming a member of, I do know that they’re enthusiastic about different stakeholders as effectively. However, however what I’m anticipating to see is sort of transferring that collaborative framework ahead and looking out in direction of which areas of labor they wish to undertake first, and the place they will they will already discover settlement, and the place there seems to be tractable work, and what issues they could save for later.

Ronan Murphy
And Invoice simply from, from the European perspective, perhaps not the recent matter in Brussels proper now, there’s a lot taking place elsewhere, however perhaps set the scene on what, what we really feel may come from Brussels, what they could contribute or not at this level in these discussions.

Invoice Echikson
Yeah, I imply this, fewer right here in Brussels are speaking in regards to the AI Safety Summit in San Francisco, which is a half a world away. I believe Europeans have lengthy felt that they had been, and are, the leaders in pursuing regulation to make sure the protection. They handed an AI Act, the primary type of binding authorized restrictions on AI, and at this level, I believe they’re having just a few second ideas about it, as a result of they’re most involved, the discuss right here is generally targeted on competitiveness and ensuring, already, persons are saying that the Europeans have missed the AI revolution, and it’s, they’re too far behind to meet up with the US or Chinese language lead. However, so within the latest hearings that we had for the brand new commissioners, there was actually a way that they had been going to to roll again, or not less than cut back among the restrictions, notably for startups, and that they’ll be speaking slightly bit much less about security and restrictions on AI and slightly extra on innovation and inspiring AI within the coming years.

Ronan Murphy
Yeah, and I believe perhaps the the Terminator imaginative and prescient of what’s going to occur with AI has has receded considerably. You talked about, , the the EU is likely to be, Europe is likely to be perceived as falling behind, you talked about China. Possibly you converse to that for a second, Heather, I imply, China’s engagement right here is, is restricted, and invites are most likely restricted. Is that going to undermine or is that all the course of, or is {that a} good a superb start line that simply can’t be trusted? Is that the considering from, from, yeah, from the US-EU perspective. How do you suppose that’s going to form out?

Heather West
I believe there’s some hesitancy. However I really, I believe that beginning with a smaller group of aligned international locations makes numerous sense. It’s actually laborious to return to broad consensus when, when everybody’s ranging from drastically totally different locations and drastically totally different understandings of what AI safety or AI security may imply. Beginning small after which rising that group makes numerous sense to me. And discovering alignment earlier than you exit into the larger, broader world, and maintaining that invitation listing slightly bit smaller might be a reasonably productive technique.

Ronan Murphy
Okay, so, yeah, I imply, we’ve seen one other [unintelligible] and so forth. Sorry, Invoice, I don’t know when you have, something wish to add there? I heard a voice, might need been yours.

Invoice Echikson
No, no, I’m positive about it. Once more, I believe, , security is slightly bit,simply to comply with up, fears are actually decreased slightly bit. I believe this was actually when Open AI launched the chat bot, that basically individuals mentioned, wow, this might actually do one thing after which it turned a really political problem in Europe. And now, once more, it’s among the fears have been decreased and among the worries are extra about, hey, we want this too, we actually must catch up.

Ronan Murphy
Yeah.

Heather West
And there’s worth in having a gaggle of nations who’re actually trying to make sure innovation inside their framework and their context right here. In the event that they’re enthusiastic about competitors with China, which I don’t wish to say that I do know precisely what they’re considering, however I’m positive that that is part of the dialog, and attempting to ensure that now we have a framework and a shared framework to maneuver AI innovation ahead rapidly.

Ronan Murphy
Okay, so I’d encourage anybody on the decision, if you wish to increase fingers, when you have a query to ask, please achieve this. I do know we’ve we’ve allotted an hour of time, however that’s not a goal. So we had been comfortable to maneuver issues faster than that. And I believe that political momentum that was maybe there at the beginning, there was an curiosity, as a result of it appeared that ChatGPT got here out of the blue for lots of people, such as you talked about Invoice. Is the political urge for food, has it dissipated? Are they comfortable at hand over? Such as you’ve talked about, Heather, that is the technical assembly, it’s at a technical stage. Do you suppose that that now we have moved to that time now that, and that’s most likely excellent news? Rapidly, and I do know we [unintelligible].

Invoice Echikson
I’d say, I can reply. You already know that initially, when the AI Act was proposed right here in Europe, it was a technical dialogue, it was fairly benign. It was actually slightly extra transparency, very slender restrictions after which that modified because it turned such a entrance web page problem and AI started to dominate everybody’s discussions. Possibly because it recedes and the emphasis adjustments, I believe that, , the legislation is there and the way will probably be applied can be an actual, is what’s preoccupying European coverage makers, and the way they put in guidelines, for instance, on copyright goes to be an enormous problem, as a result of they’ve very restrictive guidelines right here in Europe. However once more, they’re going to attempt, I believe, not less than, to weaken or dilute the legislation sufficient in order that it permits innovation.

Ronan Murphy
Okay. Okay, I believe Matt, Matt, you’ve received your hand up if you wish to ask your query, please.

Matt O’Brien
Thanks. I believe Rick has the same query, however the Biden administration is sort of internet hosting this occasion. President Elect Donald Trump has mentioned he desires to repeal Biden’s signature AI coverage. What does that imply for the entire different international locations which might be attending? I imply, or do you suppose they’re going to be searching for some sign as to if there’s going to be adjustments? Do you suppose that can change how international locations are collaborating on these things?

Heather West
I believe, I believe that this, this can be a good demonstration of why I’m excited that it’s a really technical group coming collectively. The technical items aren’t going to vary that a lot with coverage, with a change of administration. However additionally it is essential to do not forget that the Biden AI EO is one step in an extended course of. We shouldn’t neglect that the 2019, Trump administration AI EO. targeted on security, safety, and, I consider resiliency, and they also’re very comparable themes, and there’s no purpose to consider that we’ll be doing the 180 in relation to the work of the AI Safety Institute. Though, I don’t suppose that anybody is aware of precisely what the change in administration means for the AI Safety Institute on this work.

Invoice Echikson
Yeah, and I believe from Europe, that is bigger. This isn’t the entrance line problem that Europeans are nervous about proper now, it’s extra about the specter of tariffs, notably tariffs on German automobiles, for instance, and whether or not they must retaliate, or whether or not they can diffuse that blooming commerce battle that’s on the prime of the financial agenda. On AI, I believe the Europeans and People, whereas sharing greater than we’d have anticipated, , there was extra of a political consensus right here that regulation was required. It’s finished and dusted. And, , I don’t suppose the Europeans will, this would be the focus of any type of Transatlantic Trump infused debate.

Ronan Murphy
Okay. Okay, I believe we, Rick, I believe we, we had been talking to your query there, I don’t know if there’s something you wish to add there. Do you hear us Rick?

Heather West
I believe, I believe I can add one further piece right here. After we’re enthusiastic about, very particularly, the way forward for the AI Safety Institute, the US AI Safety Institute, it’s helpful to do not forget that there’s very broad trade assist for its work. I consider there are a number of 100 members of the AI Safety Institute Consortium. And in my work, I discuss with a few of these corporations regularly, and I hear numerous assist. And so I believe, I believe any, any new administration, as they consider what they wish to do on AI coverage goes to have to consider that.

Ronan Murphy
Okay. Okay, Rick, I believe you muted your self from unmuted.

Rick Weber
Are you able to hear me?

Ronan Murphy
I can now, yeah.

Rick Weber
Okay, sure. So Mike, I suppose, I suppose most of my query was answered, however I suppose I’m nonetheless questioning if the E, if the EO is repealed, does all people anticipate that the AI Institute in the US and in different international locations will proceed to have the assist that it has now?

Ronan Murphy
Go on, Heather.

Heather West
Sure, sure. So actually altering or repealing the Biden AI EO wouldn’t affect the work of different international locations. And, and it’s, it’s, , the AI Safety Institute isn’t talked about within the Biden AI EO. It’s an outgrowth of the course that they had been taking. And it, we don’t know. Nobody is aware of precisely what this type of said repeal of the AI, the Biden AI EO may seem like, however what I’m listening to is it’s more likely that there can be items of it which might be drawn again and doubtlessly some alternative language, however that the general course might not change. The emphasis is more likely to change.

Rick Weber
Okay, can all people nonetheless hear me?

Ronan Murphy
We will. Yeah, go forward.

Rick Weber
Okay, so then a comply with up query by way of end result of the assembly subsequent week. It’s targeted on technical so what are you anticipating can be product that comes out of the assembly subsequent week?

Heather West
I hope that there’s a piece plan.

Ronan Murphy
So we had a declaration from [unintelligible], I believe, at the beginning, however that was at a political stage. And so yeah, work plan could be attention-grabbing, if I would, it’s associated to what Rick simply requested. Is open supply versus not open supply going to be mentioned? Is that the extent that they’ll get to, or is that too commercially delicate for for this group do you suppose?

Heather West
I don’t understand how explicitly will probably be talked about, however actually will probably be a part of discussions given, given the predominance of varied sorts of supply mechanisms and enterprise fashions behind AI techniques. And, and there’s numerous dialogue about what the definition of open supply is. I don’t suppose that can essentially be a subject of dialog, however I, it could be a mistake for the AI security institutes to disregard that, that there’s a variety of fashions behind these AI fashions to concentrate to as they do their work.

Ronan Murphy
Yeah, yeah. And I believe Invoice, we’ve heard from individuals in Brussels and elsewhere inside Europe that the common, the AI Act, because it stands, can not get in the way in which of doing enterprise. And as you mentioned, that they’ve rolled again slightly, and it appears unlikely the European Union goes to be advocating for something that’s going to contradict that view,

Invoice Echikson
Yeah, I believe one of many issues that is likely to be a divergence between the US and Europe on this. I imply, on the one hand, the US has managed to type of limit the sale of {hardware} required for AI to China, forcing the Netherlands ASML to not promote its most superior machines to to China. However I haven’t heard any dialogue like I’ve within the US about type of limiting the export or diffusion of AI type software program to China. I believe that’s a step too far for many Europeans and wouldn’t get a lot assist, particularly in the event that they felt they had been being buoyed to forestall the collaboration with the Chinese language over a few of this AI know-how.

Ronan Murphy
Yeah, and for these , we do have a few good items on export controls and their affect on on LLMs and AI, AI software program, Invoice alluded to on cepa.org. And I’ll ask the group, is there anybody else who want to put their hand as much as ask something particular? Craig, I believe you’re a latest addition and please go forward. Within the meantime, are there any concrete tasks, issues that we may hope to, , this group may concentrate on that might they might ship one thing on issues that make sense to the remainder of the world. Possibly Heather is one of the simplest ways placing it.

Heather West
Positive. No I, one of many issues that I hope the AI Safety Institutes take note of is like, the place they will discover this, this massive consensus already, one of many issues that I believe most of those international locations, if not all, agree on typically, is the significance of cybersecurity and the right way to safe software program techniques and a few of these greatest practices. However they, they’ve the chance to actually translate that into the AI context, into that ecosystem, and to speak about the place this is likely to be totally different, or the place there is likely to be gaps, and that’s one thing that’s actually achievable and concrete and measurable, such that they might make some actually fast progress over the following six months.

Ronan Murphy
Okay, so one thing concrete may come from this, hopefully. Okay. Ruth.

Ruth Reader
Hello, I’m curious, AI, and perhaps I’m getting forward of type of cheap progress, however AI is such a broad know-how, so like within the context of security, there actually goes to require some trade particular safeguards I suppose. Is that part of discussions already? Can we anticipate that to be part of discussions? Any type of perception there?

Heather West
I believe that’s already part of discussions and there’s, there’s dialogue about whether or not you’ll be able to create excessive stage frameworks and excessive stage greatest practices that apply throughout the board. However I believe there’s additionally broad settlement that making use of them contextually and inside a specific sector or explicit use case will at all times require further translation. Simply because, , using these AI techniques in healthcare and finance is totally different than, , me enjoying round with the chat bot. You already know, that these have very totally different security implications. However I do suppose that there’s, there’s already dialogue about sort of the way you measure these security implications and the way you measure these dangers that I’m actually excited to see what comes of it.

Invoice Echikson
Yeah, and in Europe, undoubtedly, within the AI Act, that was one of many key debates, was how, what must be seen as critically or harmful and subsequently, or excessive danger I believe they known as it, sorry, within the act. And it began out as a really slender listing, perhaps, principally about job looking and perhaps well being, after which as as they received scared and frightened it was expanded, and now I believe they’re attempting to slender it once more slightly bit. So, it undoubtedly was on the coronary heart of the controversy was, how do you outline which components of AI or which machine studying was in scope, and in scope, in what method, with kind of restrictions.

Ruth Reader
Simply to comply with up actually rapidly. Are you able to each discuss slightly bit in regards to the metrics? Like, what sort of metrics are we speaking about up to now by way of, like, how we assess security.

Heather West
Invoice do you wish to go first? Or ought to I?

Invoice Echikson
I believe you’ll must take that one. I’m not fairly positive how, if I’ve the detailed information of the AI Act for what metrics. I can say basically, one of many criticisms I’m listening to increasingly in Europe is that they weren’t doing affect assessments, that they weren’t having metrics after they went forward with sure rules right here, particularly numerous the digital rules. And I believe one of many issues that perhaps modified is there’s a requirement amongst Europeans themselves, the governments, that the European Fee simply can’t type of transfer ahead with out, with out actually far more research in regards to the potential affect. These affect assessments now have been type of like field checked, , and probably not reviewed independently, simply reviewed by the fee itself. And I’ve heard numerous discuss altering that going ahead. Heather?

Heather West
Metrics are laborious, proper? I, and that’s, however that’s, within the US, that’s what NIST is actually, actually good at they usually’re, after all, operating the AI Safety Institute. And I believe they’re going to be taking a look at, at among the issues which you could measure, and beginning with the issues which you could measure simply. You already know accuracy, in a specific use case that requires accuracy, how do you measure accuracy while you’re speaking about doubtlessly generative AI. Whether or not that’s summarization or transcription, there’s numerous issues that is likely to be in scope there. I believe that we’re going to begin to see higher metrics round resiliency and robustness. These are difficult measures, and I’m undecided that we’re ever going to have a very easy algorithm for them, however I believe that we’re going to have a lot better measures for that analysis. And to Invoice’s level, I believe we’re going to see a much wider and effectively understood course of round issues like AI affect assessments. You already know, listed here are the issues that you need to suppose via, right here is the method to do this danger analysis, even when we will’t provide you with the numbers and we don’t have essentially a concrete, numerical scoring metric. We most likely can have extra, extra versatile methods to measure and perceive and talk the danger of a given system.

Invoice Echikson
And simply so as to add that Europe’s AI Act insists on sure transparency necessities, handing over information to be independently reviewed by this AI workplace. I imply, the AI workplace was simply began, like, a month in the past, and it’s actually, I believe they’ve 100 staff, they’re going to go as much as 200, in line with the brand new commissioner. However it’s been, it’s too early to actually, they don’t have the information to make any of these type of detailed affect assessments proper now, or to place even, to grasp the metrics that you simply’re speaking about. Possibly they’ll get there, perhaps Europe will get it. Proper now, the opposite factor I ought to say is numerous the AI merchandise aren’t being rolled out in Europe due to uncertainty about this. We wrote an article about it. If you happen to look all the key AI, US AI corporations have both, , not rolled out in any respect or slowed down their rollout. It’s not simply the AI act, it’s additionally privateness guidelines and the antitrust guidelines which have, type of a mix that has made regulatory uncertainty their type of watch phrase. Ultimately, I believe they’ll roll it out, it’s too huge a market, and I believe the dangers aren’t as excessive as they’re perhaps saying, there may very well be some political factors in that, however, however we’re seeing a slower roll out right here of AI. Yeah, and I as soon as did a, , you do a Google search in Europe and a Google search on the identical time, we did it in Washington, and it got here up with, , the Washington one was simply an AI consequence and right here in Europe it was 10 hyperlinks, it was actually a lot worse. So.

Ronan Murphy
Yeah, so already having an affect. And, such as you say, Invoice, it isn’t essentially even the AI Act, the GDPR has undoubtedly had an affect on

Invoice Echikson
I imply, I’ve been asking questions left and proper about which components of the rules are inflicting this refusal to launch the merchandise in Europe. And, , I get it. It’s a mix.

Ronan Murphy
Yeah. And Michael has kindly shared the, helpfully shared as effectively, the hyperlinks to a number of of these articles within the chat, for many who have an interest and prefer to learn a bit extra. Ruth, is, is there anything I don’t know out of your perspective? Was there a, what metrics had you in thoughts? Or is that this a?

Ruth Reader
Positive. Nicely, I, I’m a well being and know-how reporter, so I’m actually trying loads at how the US is dealing with AI within the healthcare house. And it’s attention-grabbing, you’re seeing numerous like trade led efforts to attempt to deal with AI. Clearly, HHS has some type of varied steering that they’ve put out round AI’s use in medication, relying on the context.However it’s nonetheless very, it’s not, we don’t have numerous metrics both, metrics are an enormous dialog, about the way you type of even perceive these applied sciences and in addition the context through which they’re examined. I imply, medication is type of distinctive in a sure method, as a result of, , an enormous piece of the dialog that persons are having is, the place can we even take a look at it, proper? And numerous feeling is that you simply sort of have to check the AI situationally. And that has numerous and that’s difficult. So anyway, that’s why I used to be type of.

Invoice Echikson
Very delicate. I imply, well being information was at all times thought of type of the excessive danger kind of factor, I believe even within the early drafts, as we received. However I do suppose, additionally it was attention-grabbing, I used to be speaking to, apparently, there’s Swedish startups which might be doing Swedish well being information so that you simply, and constructing it on prime of ChatGPT OpenAI know-how, in order that , you possibly can see this localized as as effectively. And so they, as a startup, they’d not be topic to most of the restrictions.

Ronan Murphy
Sure, it can assist. As an trade that’s accustomed to regulation, expects regulation. And so, yeah, perhaps, perhaps there’s a pure hyperlink. Nicely, inform us the principles, we received’t do something with out the principles. I don’t know, perhaps there’s a tradition of that and that can play a job. And so, yeah, I imply, sorry, Heather, I don’t know if you happen to had been coming in.

Heather West
No, no. I used to be simply gonna say I believe there are methods that they will sort of up stage the situational metrics slightly bit. Healthcare is a improbable instance really. You already know, you’ll be able to discuss ranges of accuracy and precision, and in some contexts you’re going to want a low stage of accuracy and precision, and in lots of contexts, within the healthcare world, you’re going to want very excessive ones. You may, you’ll be able to most likely put some preliminary scores towards a given mannequin or software, however you’re nonetheless going to have to guage them in context for that closing evaluation.

Ronan Murphy
Yeah, and I, now {that a} query that comes up generally is, you’ve talked about, Invoice, Swedish, Swedish well being information might be in an excellent state, if you’ll, and it’s most likely effectively maintained. It’s a, it’s a closely digitalized society, individuals use the well being service in that method. That’s not essentially the case elsewhere, and that in of itself, is a danger. If the information isn’t prepared to be used, then you’ll be able to’t plug it into mentioned mannequin and anticipate outcomes that matter or which might be correct or useful or secure, doubtlessly.

Invoice Echikson
I believe the case in Sweden, or right here, I imply, there’s a centralized information system, the Nationwide Well being Providers, however native corporations have extra entry. So, I imply, , they will prepare their information on the native information, the place they’ve extra of it than the worldwide, normally American corporations.

Ronan Murphy
Which, which sounds an terrible lot just like the dreaded sovereignty phrase, and so an area, an area supplier, can entry the native information.

Invoice Echikson
Sure, however I believe that is extra like they’ve a aggressive benefit as a result of they’re native.

Ronan Murphy
Yeah, okay, yeah.

Invoice Echikson
I don’t suppose it’s the federal government requiring or limiting or essentially, as from what I’ve heard, even investing on this, though we’re listening to that there can be a brand new, , EU AI Act just like the CHIPS Act, the place they’ll put public funds into AI catch up. That’s most likely one of many first issues that can come out from, from Europe within the new administration there.

Ronan Murphy
Yeah, and also you talked about HHS route, and we’ve coated the Trump administration’s coming in. There’s numerous underneath new administration indicators hanging on doorways across the place, and we can not truthfully reply what the affect goes to be but. And I raised sovereignty, Heather, as a result of it has been, it’s mentioned, it’s a really huge matter in, in Europe, has been capable of stand on our personal two ft, no matter, no matter definition you need on it. Are sovereignty and security appropriate, even inside a small group like this? If, are they, can you have got each on the identical time?

Heather West
Completely.

Ronan Murphy
Yeah, okay.

Heather West
I believe which you could have, effectively, let me again up slightly bit. There have been, there’s been some actually nice work, sort of bringing collectively the frequent themes of AI security throughout borders and throughout frameworks and throughout cultures. And largely, we agree. We’d, at a nationwide stage, at a cultural stage, put emphasis on various things, however typically it’s the identical, the identical classes. There’s no purpose that everybody must do the very same factor. It ought to match for his or her their firm, or their nation, their corporations, their tradition, however, however, like, 95% of it’s most likely going to be the identical. And so, so there isn’t any purpose that all of us must be an identical. Nonetheless, I believe it’s actually essential that we’re all appropriate.

Ronan Murphy
Yeah, okay. So yeah, interoperability at some stage.

Heather West
It’s a really related world we’re in.

Ronan Murphy
Yeah, okay. So once more, encourage anybody else with any particular questions. If not, I imply, we’re, I’m comfortable to wrap up with any closing remarks. Invoice or Heather, you’d like so as to add, that’s nice, but when not, Michael, perhaps you, if there’s something you wish to say on the logistics right here, or the dissemination of the decision from right this moment, please go forward and earlier than you do, simply say because of everybody for becoming a member of, and notably because of Heather and Invoice on your contributions. Very a lot admire it.

Michael Newton
Thanks very a lot, Ronan and thanks Heather, thanks Invoice, and thanks everybody for becoming a member of the query, becoming a member of the dialog. We can be emailing out a hyperlink to the recording of this video, and we’ll even have a tough, auto generated transcript. This can be adopted by an official transcript that can go on our web site within the coming days. You probably have every other inquiries, please be happy to e-mail press@cepa.org, and we’ll be capable of join you both with Heather or Invoice if you happen to’ve received comply with up questions or any of our different consultants. Thanks very a lot, and have a fantastic day. Bye.